Ketogenic Diet

Diet Reference Archive: a medical treatment and a popular diet

Ketogenic Diet

A way of eating so low in carbohydrate and so high in fat that the body switches to burning ketones, made from fat, for much of its fuel. It is a prescribed medical treatment for epilepsy that has not responded to medication, and, separately, a popular diet followed for weight loss, blood sugar and general health.

Family: Low-carbohydrate and high-protein diets Promoted by: epilepsy treatment centers for the medical diet; authors, companies and online communities for the popular diet First appeared: 1921, as an epilepsy treatment Medical use. GRADE: Low for a 50 percent or greater reduction in seizures in children with drug-resistant epilepsy, ketogenic diets against usual care, at 3 to 4 months; published Cochrane review (Martin-McGill et al., 2020). Popular use. GRADE: Moderate for probably little to no difference in body weight in adults with overweight or obesity, low-carbohydrate diets as a group, most not ketogenic, against balanced diets, at 1 to 2 years; published Cochrane review (Naude et al., 2022)

ON THIS PAGE

1. In brief
2. Overview
3. In practice
4. For the coach
5. The evidence (opens on request)
6. Mechanisms (opens on request)
7. Risks, contraindications and myths (opens on request)
8. Key terms
9. Review questions
10. References
11. Evidence rating and review record

The summary and sections 1 to 3 are the short version and are open when the page loads. Sections 4 to 6 hold the research detail, the biology, and the cautions. Each of those opens when you select its heading, and you can read the page without opening them. Back to the Diet Reference Archive index.

In brief

What is it? A way of eating that keeps carbohydrate so low, often to about 20 to 50 grams a day, that the liver starts turning fat into substances called ketones, which the body and brain then use for fuel. The word "keto" covers 2 very different things. One is a medical treatment for epilepsy, prescribed and run by a neurology team with a registered dietitian. The other is a popular diet followed for weight loss, blood sugar or general health, often without any professional involved.

Does it work? For epilepsy, yes, for some people: in randomized trials of children whose seizures did not respond to medication, about 45 in 100 on a ketogenic diet had their seizures cut by at least half after 3 to 4 months, compared with about 8 in 100 on usual care. That evidence is low in certainty, and the evidence in adults is weaker.

For the popular uses, in adults with type 2 diabetes, trials pooling low and very low carbohydrate diets found lower HbA1c, a measure of average blood sugar, and lower triglycerides at 6 months. The HbA1c difference was smaller and the weight difference was no longer clear at 12 months, while triglycerides remained lower. These pooled findings do not isolate ketogenic diets. Over 1 to 2 years, low-carbohydrate diets as a group probably make little to no difference to body weight compared with balanced diets. No trial found for this page has tested whether a ketogenic diet followed for weight or diabetes changes heart attacks, strokes or length of life.

Is it safe? In its medical form it is a monitored treatment with known side effects, including constipation, kidney stones, high cholesterol and slower growth in children, which the treating team watches for. In its popular form the 2 cautions that matter most are these. LDL cholesterol rises in many people, sometimes steeply. And for anyone taking medicine for diabetes, a sudden drop in carbohydrate can cause low blood sugar on insulin or a sulfonylurea, and a dangerous acid buildup called ketoacidosis on the class of drugs called SGLT2 inhibitors.

Who should not do it without talking to a doctor or dietitian first? Anyone with epilepsy, whose diet therapy belongs to their neurology team. Anyone taking insulin, a sulfonylurea or an SGLT2 inhibitor, or medicine for blood pressure. Anyone who is pregnant, trying to become pregnant or breastfeeding. Children and teenagers. Anyone with kidney disease, a history of kidney stones, pancreatitis, liver disease or very high cholesterol, including inherited high cholesterol. And anyone with a history of disordered eating.

Related pages: Mediterranean Diet, Low-Carbohydrate Diets, Atkins Diet, Carnivore Diet and Intermittent Fasting. Full list on the archive index.

1. Overview

What you eat

A ketogenic diet is defined by what it does to the body rather than by a list of foods. Carbohydrate is cut so far that the liver begins making ketones from fat, a state called nutritional ketosis, and most of the day's energy then comes from fat.

The amounts differ by version. The classic medical diet for epilepsy is built on a ratio of 4 grams of fat to every 1 gram of protein and carbohydrate combined, so that about 90 percent of calories come from fat (Kossoff et al., 2018). The modified Atkins diet, a looser medical version, starts at about 10 to 15 grams of carbohydrate a day (Kossoff et al., 2018). Popular versions often aim for 20 to 50 grams of carbohydrate a day, with protein held at a moderate level and fat making up the rest, as in the well-formulated version tested in the Keto-Med trial (Gardner et al., 2022).

The foods that fit are those with little or no carbohydrate. The American Heart Association's 2023 review of popular diets lists the foods a very low carbohydrate or ketogenic pattern emphasizes as nuts and seeds, red meat, poultry, fish and seafood, eggs, full-fat dairy foods and oils, with nonstarchy vegetables and berries included in smaller amounts (Gardner et al., 2023). In a well-planned version, such as the one tested in the Keto-Med trial described in section 4, people were also asked to eat more than 3 servings a day of nonstarchy vegetables such as leafy greens, broccoli and peppers, and to keep up their intake of salt, potassium and fluid (Gardner et al., 2022).

What you avoid or limit

Almost every food high in carbohydrate is left out. In the American Heart Association's description, the pattern avoids fruits other than berries, all grains, legumes such as beans and lentils, and added sugars (Gardner et al., 2023). In everyday terms that means bread, pasta, rice, cereal, most fruit, beans and lentils, sweets, juice and regular soda.

In the medical diets the limit is a prescription, with foods weighed and the menu calculated by a dietitian (Kossoff et al., 2018). In popular versions people set their own limits, and trials show that carbohydrate intake often drifts above ketogenic levels over the months of a trial (Li et al., 2023; Parry-Strong et al., 2022).

Where it came from

The diet began as a medical treatment. Fasting and other dietary regimens had been used to treat epilepsy since at least 500 BC (Wheless, 2008). In July 1921 a physician named Wilder published 2 short papers on consecutive days in the Mayo Clinic Bulletin: the first proposed that a diet producing ketosis might reproduce the benefit of fasting, and the second reported its use in 3 patients with epilepsy (Auvin, 2021). The diet was widely used for about 2 decades, then declined as antiseizure medicines arrived, until by the end of the twentieth century only a small number of children's hospitals offered it. Its use then grew again from the 1990s onward (Wheless, 2008), and a randomized trial in children was published in 2008 (Neal et al., 2008).

The popular diet has a separate history. Low-carbohydrate weight-loss plans such as the Atkins diet are older than the current keto trend, and the American Heart Association groups Atkins, ketogenic and well-formulated ketogenic diets together as very low carbohydrate patterns (Gardner et al., 2023). In recent years "keto" has become a name in its own right, with its own products and a large presence on social media. In one 2024 study, an Instagram account set up to look like a teenager's found about 19 million posts tagged #keto, rising to 21.5 million at a second check (Lister et al., 2024).

What it claims to do

For epilepsy, the claim made by the clinicians who prescribe it is specific and limited: that a ketogenic diet can reduce seizures in some people whose seizures do not respond to medication, and that for 2 rare disorders of brain energy metabolism it is the treatment of first choice (Kossoff et al., 2018).

The popular claims are broader. Virta Health, a company that delivers a ketogenic-style program through an app and remote clinicians, tells prospective members that "by changing what you eat, you can help your body shift from using sugar to burning fat for fuel," that members "enjoy sustained weight loss and type 2 diabetes reversal," and that its program is linked to a 56 percent reduction in heart attack, stroke and death, a figure it attributes to a retrospective claims-based study listed as accepted for publication with the citation pending (Virta Health, n.d.). Claims a client may meet online go further still: that keto burns fat in a way other diets cannot, cures diabetes, or starves cancer. Section 4 takes the evidence on each kind of claim in turn, and section 6 takes several of them as myths.

What it gets right

The ketogenic diet removes most of the foods that general heart-health guidance also asks people to limit: added sugars, sugary drinks, refined grains and sweets. The American Heart Association's review credits very low carbohydrate patterns with an emphasis on nonstarchy vegetables, nuts and fish and minimal added sugar (Gardner et al., 2023). Cutting carbohydrate also lowers blood sugar quickly, which is why many people with type 2 diabetes see their numbers improve within the first months (Goldenberg et al., 2021).

On the medical side, the ketogenic diet has randomized trial evidence as a treatment for a neurological condition. For children with drug-resistant epilepsy it is a recognized treatment, backed by an international consensus on how to run it safely (Kossoff et al., 2018; Martin-McGill et al., 2020). A coach can honor a client's interest in either use without endorsing the whole package.

How clients describe it, and what it gets confused with

Clients say "I'm doing keto," "I'm cutting carbs," "I'm in ketosis," "I'm fat-adapted," "I do lazy keto" or "I'm keto most of the time," or they bring a product: keto bars, keto bread, keto snacks, ketone drinks or test strips. Others say "my son is on the ketogenic diet for his seizures," which is a medical treatment and a different conversation. These range from a supervised therapy weighed to the gram, to a strict popular diet, to a general cut in sugar and bread, to buying products labeled keto. Asking what the word means to this client comes before anything else.

The diet is most often confused with the broader family of low-carbohydrate diets. The boundary is the depth of the cut: a ketogenic diet restricts carbohydrate far enough to produce ketosis, usually to 50 grams a day or less, while the low-carbohydrate diets in the research usually allow more, up to about 130 to 150 grams a day (Goldenberg et al., 2021; Naude et al., 2022). That family is described on the Low-Carbohydrate Diets page of this archive, listed on the archive index.

The diet is also confused with the Atkins diet, whose first phase is ketogenic; with the carnivore diet, which removes plant foods altogether; and with Intermittent Fasting, because both fasting and keto raise ketones. Each has its own page in this archive. And it is confused with very low calorie ketogenic diets, which restrict total calories as well as carbohydrate, and with exogenous ketones, which are supplements sold to raise blood ketones without a change in diet (Kossoff et al., 2018).

2. In practice

A typical day of eating

The example below is a popular, home-cooked version, written to match the Keto-Med trial's well-formulated diet rather than the medical diet, whose menus a dietitian calculates individually (Gardner et al., 2022; Kossoff et al., 2018). Breakfast is 2 or 3 eggs scrambled in olive oil with spinach and mushrooms, and coffee with cream. Lunch is a large salad of leafy greens, cucumber, avocado and grilled chicken with an olive oil dressing, and a small handful of almonds. An afternoon snack is cheese with celery, or a few berries with full-fat Greek yogurt. Dinner is baked salmon with roasted broccoli and cauliflower and a side of green beans. Water, unsweetened tea and broth are the usual drinks.

There is no bread, rice, pasta, potatoes, beans, fruit beyond a few berries, or sweets. The same day could also be built mostly from processed meats, cheese and packaged keto snacks, and the 2 versions differ a great deal even though both count as keto.

Cost, products, and who profits

What the popular diet does to a grocery bill depends on the foods chosen, and the search for this page found no study of what it costs American households. The closest was an Australian study that priced 7-day meal plans for adults with epilepsy. For women, the classic ketogenic menu, set at a ratio of 2.5 to 1, cost slightly less than a typical Australian diet, 70.07 against 75.91 Australian dollars a week, while for men every ketogenic menu cost more; moving from a typical diet to the modified ketogenic menu raised weekly food costs by 13 percent for women and 44 percent for men (King et al., 2026). Those are modeled menus at Australian prices, not what people paid.

Money attaches to the name in several places. In a study of 757 gluten-free bakery products labeled ketogenic or low-carb, some did not meet any definition of a ketogenic diet, so the label alone does not tell a buyer what is inside (Gasparre et al., 2022). Ketone meters and test strips are sold for home use, and exogenous ketone supplements are widely available, although the international epilepsy consensus states there is no evidence for their role in treating epilepsy (Kossoff et al., 2018). Companies sell ketogenic programs through apps and remote clinicians; Virta Health is one, and 7 of the 9 authors of the 2-year study of its program listed the company as an affiliation (Athinarayanan et al., 2019; Virta Health, n.d.).

On the medical side, several authors of the international epilepsy consensus disclose ties to companies that make ketogenic formulas and, in one case, to Atkins Nutritionals (Kossoff et al., 2018). None of these ties makes the research wrong. It is worth knowing who has a stake in it. A person with limited money can build a ketogenic diet on eggs, canned fish, frozen vegetables, cabbage and cooking oil, but the diet removes rice, beans, oats and potatoes, staples that many households on a tight budget rely on.

Environmental footprint

The footprint depends mostly on which foods a person chooses within the diet, and a version built mainly on meat and dairy carries more than one built on plants. In a national U.S. survey, 77 adults whose single recorded day of eating fit a ketogenic definition, 50 grams of net carbohydrate or less, had the highest average estimated greenhouse gas emissions of 6 diet types: 2.91 kilograms of carbon dioxide equivalents per 1,000 calories, against 0.69 for vegan days and 2.23 for typical omnivore days, although the difference from omnivore days was not statistically significant (O'Malley et al., 2023). The figures were estimated from one day of reported eating, not measured. The American Heart Association notes that the patterns scoring highest in its review also have a lower environmental impact because they include less meat (Gardner et al., 2023).

Cultural fit

The diet fits most easily into food traditions built around meat, fish, eggs, dairy and nonstarchy vegetables, and it conflicts with the many traditions whose daily food centers on rice, corn tortillas, bread, noodles, beans, lentils, plantains or potatoes. Following it strictly can mean eating differently from the rest of the family at most meals. In a study of university students in Brazil, where rice is a staple, students following a low-carbohydrate diet reported more binge eating and food craving over 3 months than other students, and the authors suggested that giving up culturally meaningful foods such as rice may contribute (de Oliveira et al., 2025); that is their interpretation of survey data, not a tested cause. In the U.S. survey above, Hispanic and Black respondents were more likely than White respondents to have a day of eating that fit the keto definition (O'Malley et al., 2023).

What makes people stop

Keeping to the diet is a central practical problem in both uses. In the epilepsy trials, people left because their seizures did not improve or because they could not tolerate the diet (Martin-McGill et al., 2020). Adult epilepsy centers named an inadequate response and difficulty coping with the carbohydrate restriction as the main reasons, followed by food preferences, time, limited variety, cost and social occasions (Cervenka et al., 2021).

In the popular use, a review of 11 studies in type 2 diabetes and excess weight found that participants generally ate more carbohydrate and less fat than the diet called for (Li et al., 2023), and dropout from the ketogenic arms of trials reached 54 percent in one review (Parry-Strong et al., 2022). In the Keto-Med trial, adherence was higher while meals were delivered than when participants bought and cooked their own food (Gardner et al., 2022). The American Heart Association lists the "keto flu" of the first days as a possible deterrent and judges that the diet's restrictiveness likely limits long-term adherence (Gardner et al., 2023).

What happens when you stop

For epilepsy, the international consensus advises considering stopping the diet after about 3 months if it has not helped and at about 2 years if it has. About 80 percent of children who are seizure-free on the diet stay seizure-free after it ends, and the diet is usually tapered over 1 to 3 months, because seizures can worsen during weaning (Kossoff et al., 2018).

For the popular uses, the early differences fade. In the 2021 review of trials in type 2 diabetes, the weight difference at 6 months was gone by 12 months and the HbA1c difference was about half as large, while lower triglycerides remained (Goldenberg et al., 2021). In the Keto-Med trial, during 12 weeks of free choice after both diet phases, participants' eating looked more like the Mediterranean-style diet than the ketogenic one; their blood sugar and weight stayed below where they started, and their LDL cholesterol stayed above it (Gardner et al., 2022).

Common variations

The medical diets come in 4 main forms: the classic diet, built on the 4 to 1 ratio; the medium-chain triglyceride diet, which uses a type of fat that produces more ketones and so allows a little more carbohydrate and protein; the modified Atkins diet, starting at about 10 to 15 grams of carbohydrate a day without weighing foods; and the low glycemic index treatment. The last 2 are started outside the hospital, and the evidence for them has grown especially for teenagers (Kossoff et al., 2018).

Popular variations a client may name include the well-formulated ketogenic diet, which emphasizes vegetables and quality fats (Gardner et al., 2022); very low calorie ketogenic diets, which also cut total energy sharply; keto built mainly from packaged and processed foods; and patterns that cycle in and out of ketosis. The Atkins Diet, Carnivore Diet and Low-Carbohydrate Diets pages, listed on the archive index, cover those 2 diets and how the members of the family differ.

3. For the coach

SCOPE OF PRACTICE

A health and wellness coach does not prescribe, adjust, endorse or discourage a ketogenic diet, and does not diagnose or treat. The coach helps the client understand what they have chosen, notice how it is going, and decide what to do next. A ketogenic diet for epilepsy, or for any other diagnosed condition, is a prescribed treatment designed and monitored by the treating physician and a registered dietitian; the coach supports the person and the family in living with it and does not change or comment on the prescription.

In the popular use, a clinician needs to be involved for any client taking insulin, a sulfonylurea, an SGLT2 inhibitor or blood pressure medicine; for pregnancy, trying to conceive and breastfeeding; for children and teenagers; and for kidney disease, kidney stones, pancreatitis, liver disease, very high cholesterol or a history of disordered eating. For a client in any of those groups, where documented harms exist, the coach's role includes making sure the client has medical oversight, and it does not include supporting the diet itself.

Who does what

The physician

For epilepsy, the neurologist decides whether a ketogenic diet is appropriate, screens for the rare metabolic disorders that rule it out, orders and reads the monitoring labs, adjusts antiseizure medicine, and decides when the diet stops (Cervenka et al., 2021; Kossoff et al., 2018). For the popular use, the prescriber adjusts insulin, sulfonylureas, SGLT2 inhibitors and blood pressure medicine before or during a large drop in carbohydrate, and orders and interprets the lipid panel and kidney tests that matter for this diet (American Diabetes Association Professional Practice Committee for Diabetes, 2026).

The registered dietitian

For epilepsy, the dietitian calculates the prescription, teaches the family to weigh and prepare meals, chooses the supplements the diet requires, and adjusts the ratio over time (Kossoff et al., 2018). For the popular use, a dietitian can build an individual plan that meets nutrient needs at the client's chosen carbohydrate level. The American Diabetes Association recommends referral to a registered dietitian for individualized nutrition therapy for people with diabetes or prediabetes (American Diabetes Association Professional Practice Committee for Diabetes, 2026).

The health and wellness coach

Helps the client say what they want from the diet and why now, and whether "keto" means a strict plan, a cut in sugar and bread, or a set of products. Explores how the diet fits the client's budget, family meals, culture and social life, and helps the client plan the habits that make their chosen approach workable. With the client's permission, can share general guidance from recognized authorities, such as the American Diabetes Association's statement on very low carbohydrate eating and SGLT2 inhibitors, and explore how the client wants to use it. Notices whether eating is widening or narrowing, and holds the referral line. For a family living with a medical ketogenic diet, the coach can help a parent or an adult patient manage the daily load of a demanding treatment: planning, stress, social events and support.

What a client bringing this diet needs

Clients bring keto for a handful of reasons: a friend lost weight on it; a diagnosis of prediabetes or type 2 diabetes and a wish to bring the numbers down without more medicine; a promise read online of fat burning or a cure; or, in a different conversation altogether, a child or an adult whose epilepsy treatment includes a prescribed ketogenic diet. The reason shapes the conversation. A client who takes an SGLT2 inhibitor and plans to start keto in a few days needs something very different from a client who wants to cut back on bread. What a coaching conversation offers that a search engine cannot is room to separate what the client wants from the plan they found, and a chance to make sure the people treating them know what they intend.

What to ask

• When you say keto, what does a usual day of eating look like for you?
• What are you hoping will be different in 3 months, and in a year?
• Who on your health care team knows you are planning this, and what did they say about your medicines?
• How does this way of eating fit with the people you usually eat with?

Referral red flags

A client who takes insulin, a sulfonylurea, an SGLT2 inhibitor or blood pressure medicine and is starting or has started a ketogenic diet goes to the prescriber before going further, for the reasons given under special populations below (American Diabetes Association Professional Practice Committee for Diabetes, 2026). A client on insulin or a sulfonylurea who describes what may be a low blood sugar episode treats the episode first, under the client's own plan, with emergency help for severe symptoms or an episode the person cannot manage; telling the prescriber is the separate step that follows. A client on an SGLT2 inhibitor who describes vomiting, heavy thirst and urination, or confusion needs urgent medical care, because ketoacidosis on these drugs can occur with near-normal blood sugar (Umpierrez et al., 2024).

Pregnancy, trying to conceive and breastfeeding go to the obstetric clinician or a dietitian, and kidney disease, past kidney stones, pancreatitis, liver disease and very high or inherited high cholesterol go to the physician before the diet starts; the reasons are set out under special populations below and in section 6. For kidney disease, individual needs depend on its stage, laboratory results and dialysis, and the kidney care team belongs in the plan. A client who mentions a large rise in cholesterol on a recent blood test is describing something for the physician to interpret.

A client whose rules keep getting stricter is worth asking about it: what each rule is for, whether it causes distress, whether it is narrowing what they eat, and whether it is getting in the way of meals with other people. There are ordinary reasons for strict carbohydrate counting, including a prescribed medical diet. Where the answers raise a real concern about disordered eating, the coach follows their own referral process, or brings the concern to the client's clinician under an existing care plan. Weight a client loses without meaning to is a reason to involve a clinician, never a sign the diet worked. In every case the coach notices and encourages the client to tell their clinician, working out the cause is the clinician's job, and none of these referrals ends the coaching relationship.

Special populations

Pregnancy and breastfeeding

The American Diabetes Association does not currently recommend very low carbohydrate eating plans during pregnancy or lactation (American Diabetes Association Professional Practice Committee for Diabetes, 2026). The search for this page found no trial of a ketogenic diet in pregnancy. It found a case series of 2 women treated with ketogenic diet therapy for epilepsy during pregnancy, whose authors conclude that safety "still has to be established" (van der Louw et al., 2017), and a survey in which 60 percent of epilepsy clinicians who had counseled patients did not recommend the diet in pregnancy (Faltersack et al., 2025). Observational studies of low-carbohydrate eating before pregnancy, not ketogenic diets, link it with lower folic acid intake and more neural tube defects, and an animal-based pattern with more gestational diabetes (Bao et al., 2014; Desrosiers et al., 2018). In breastfeeding mothers, ketoacidosis has been reported after recent dieting, most often on low-calorie, low-carbohydrate, high-protein or ketogenic diets (Al Alawi et al., 2020). Any plan to follow the diet in pregnancy or while breastfeeding belongs to the obstetric clinician and a dietitian, and for a person with epilepsy, to the neurology team as well.

Children and adolescents

For children with drug-resistant epilepsy, the ketogenic diet is an established treatment run by a pediatric neurologist and dietitian, with growth, kidney stones, cholesterol and bone health monitored; in studies longer than 6 months, the classic diet slowed growth in height (Kossoff et al., 2018). Outside that use, the American Diabetes Association does not currently recommend very low carbohydrate eating plans for children (American Diabetes Association Professional Practice Committee for Diabetes, 2026). Small, short randomized trials have also tested diets prescribed at ketogenic carbohydrate levels in children and adolescents with obesity, under close medical supervision. In one, 46 adolescents aged 12 to 18 began either a diet prescribed at 20 grams of carbohydrate a day or a low-fat diet, and 33 completed the 13 weeks (Krebs et al., 2010). In another 12-week trial, 39 adolescents were randomized to a diet prescribed at under 20 grams of carbohydrate a day for 2 weeks and then under 40 grams a day, or to a low-fat diet; the final analysis included 30 participants who reported following their diet and completed at least 8 weeks (Sondike et al., 2003). In a third, 102 children aged 7 to 12 were assigned to 1 of 3 diets for 3 months, one of which started at 20 grams a day or less and rose to at most 60 grams; the children kept to that diet less well than to the others (Kirk et al., 2012). These trials do not establish long-term safety or support unsupervised use. The search also found an uncontrolled 4-month study of 100 children with obesity, of whom 58 completed it (Paskaleva et al., 2025), and a review noting that some parents of children with type 1 diabetes are turning to the diet, often after reading about it on social media, while evidence on its safety in children is limited (Kovinthapillai et al., 2026). A coach works with the adult in front of them, and a child's eating belongs to the family, the pediatrician and, for epilepsy, the diet team.

Older adults

The search for this page found research in older adults mainly in people with mild cognitive impairment or Alzheimer's disease. A 2026 review addressed ketogenic diets and related strategies in people with cognitive impairment; only its abstract could be obtained for this page, not its full text or intervention table, so no diet-only conclusion or certainty rating is assigned here (Xi et al., 2026). For an older adult with a smaller appetite, several medicines or low bone density, a dietitian and physician belong in the plan.

Athletes and people with high activity

In a trial in elite race walkers, 3 weeks of a ketogenic diet during hard training raised fat burning but made walking at race pace more costly in oxygen; the ketogenic group did not improve its 10-kilometer time, while the groups eating more carbohydrate improved by about 5 to 7 percent (Burke et al., 2017). A review of 33 studies in adults labeled ketogenic included randomized and nonrandomized studies and some diets above ketogenic carbohydrate levels. It found no clear difference in several muscle and performance measures and a small mean reduction in fat-free mass across those mixed studies, about half a kilogram; these findings do not establish that muscle or performance is preserved during sustained ketosis (Wang et al., 2025). Sports performance questions belong to a sports dietitian.

People taking medication for diabetes, blood pressure, or blood thinning

This is the group where the popular diet most clearly needs the prescriber first: insulin and sulfonylurea doses may need to fall, SGLT2 inhibitors such as empagliflozin, dapagliflozin and canagliflozin should not be combined with very low carbohydrate eating, and blood pressure needs monitoring (American Diabetes Association Professional Practice Committee for Diabetes, 2026). Section 6 lists the interactions, including a report on warfarin. The standing rule holds: for a substantial change in eating by a client on these medicines, the client keeps the prescriber informed before the change.

People with a history of disordered eating

A ketogenic diet is built on strict rules about allowed foods, and the American Diabetes Association does not currently recommend very low carbohydrate eating plans for people with or at risk for disordered eating (American Diabetes Association Professional Practice Committee for Diabetes, 2026). In a study of Brazilian university students, those following a low-carbohydrate diet reported more binge eating over 3 months than other students, an observational finding that cannot show which came first (de Oliveira et al., 2025). Small feasibility studies are testing ketogenic diets as a supervised treatment for anorexia nervosa (Frank et al., 2026), which is clinical research inside a treatment team, not a reason to start the diet. A concern goes to the client's clinician under the existing care plan, or through the coach's own referral process.

Client misconceptions

Three misconceptions come up most: that ketosis burns body fat in a way other diets cannot, that keto reverses type 2 diabetes, and that a rise in LDL cholesterol on keto does not matter. Section 6 sets out where each came from and what the evidence shows. The practical point for the coach is that a question about what a client's own blood sugar or cholesterol result means goes to the physician.

🤝 COACHING CONNECTION

Maria is 54, has type 2 diabetes, and takes metformin and a second pill her doctor added last spring. She says her sister lost 30 pounds on keto and she plans to start Monday, "no bread, no rice, nothing." The coach asks what she is hoping will change. Maria says her energy, and getting off at least one medicine. The coach asks what her doctor knows about the plan. She says nothing yet, and she is not sure what the newer pill is called. The coach asks how she would like to find that out. Maria decides to check the bottle and call the clinic before she changes anything. Then she talks about Sunday dinners at her mother's, where rice is on the table every week, and wonders what she would do about those. Nothing was recommended, and the diet was never evaluated.

4. The evidence

(select the heading to open or close this section)

The evidence has to be read in 2 separate parts, because the medical diet and the popular diet answer different questions in different people. For drug-resistant epilepsy, randomized trials show that a ketogenic diet reduces seizures in children more often than usual care, at low to very low certainty, with weaker and less consistent evidence in adults.

For weight and type 2 diabetes, trials pooling low and very low carbohydrate diets show lower HbA1c, triglycerides and sometimes weight than comparison diets at 6 months; by 12 months the HbA1c difference is smaller and the weight difference no longer clear, while triglycerides remain lower. These pooled findings do not isolate ketogenic diets. LDL cholesterol rises on average, and no trial found for this page has measured heart attacks, strokes or deaths. Each study below says which version of the diet it tested, because "ketogenic" in one trial and "low-carbohydrate" in another can mean very different amounts of carbohydrate.

Evidence from randomized controlled trials

A trial assigns people to a diet by chance and compares them with people assigned to something else, which is the strongest way to test whether the diet itself causes a result. Ketogenic diet trials share a limitation: participants know what they are eating, so the trials cannot be blinded, and the Cochrane reviewers judged every epilepsy trial at high risk of bias for that reason (Martin-McGill et al., 2020). Many popular-use trials also lose a large share of participants, and many of the people who stay eat more carbohydrate than the diet asks for by the end (Li et al., 2023; Parry-Strong et al., 2022).

Epilepsy

The Cochrane review of ketogenic diets for drug-resistant epilepsy, updated in 2020, found 13 randomized trials with 932 participants, 711 children and 221 adults, lasting 2 to 16 months. The trials tested the classic diet, the medium-chain triglyceride diet, the modified Atkins diet and a simplified version of it. All 13 were judged at high risk of bias from lack of blinding, and the review rated the evidence for every outcome low or very low in certainty (Martin-McGill et al., 2020).

In children compared with usual care, 4 trials with 385 children reported results at 3 to 4 months. A 50 percent or greater reduction in seizures occurred in about 453 per 1,000 children on a ketogenic diet against about 78 per 1,000 on usual care, a risk ratio of 5.80, from 3.48 to 9.65, rated low in certainty. Freedom from seizures occurred in about 66 per 1,000 against 21 per 1,000, a risk ratio of 3.16, from 1.20 to 8.35, rated very low. Vomiting, constipation and diarrhea were the most common side effects, and they were also common in the usual care groups (Martin-McGill et al., 2020).

In adults, 2 trials with 141 participants compared the modified Atkins diet with usual care for 2 to 3 months. No adult in either group became seizure-free. The estimate for a 50 percent reduction in seizures favored the diet but was so imprecise, a risk ratio of 5.03, from 0.26 to 97.68, that it is compatible with no benefit, and it was rated very low in certainty. Adults on the diet may also have been more likely to drop out (Martin-McGill et al., 2020).

A trial in the United Kingdom randomized 145 children aged 2 to 16 with daily seizures who had not responded to at least 2 medicines to start the diet immediately or after a 3-month delay. After 3 months, 28 of the 73 children assigned to the diet, 38 percent, had their seizures reduced by more than half, against 4 of 72 controls, 6 percent; these are crude proportions of the children randomized (Neal et al., 2008).

Since the Cochrane review's search, a trial in 136 infants aged 1 to 24 months compared the classic diet with a further antiseizure medicine for 8 weeks and did not demonstrate a difference in seizure frequency, with a rate ratio of 1.33, from 0.84 to 2.11; it ended before all infants reached 12 months of follow-up because of slow recruitment and the end of its funding. In that trial, 3 infants died, all in the diet group, and the deaths were judged unrelated to treatment by the local investigators, a judgment confirmed by the trial's data safety monitoring committee (Schoeler et al., 2023). The journal article reporting the United Kingdom trial in children could not be obtained for this page, so its numbers above were checked against the article's abstract and against the trial's flow chart and results table in the lead author's doctoral thesis (Neal, 2008).

Cardiometabolic outcomes

Most of the trials behind the popular claims were done in adults with type 2 diabetes or prediabetes, and they differ in how much carbohydrate they allowed. A 2021 review in the BMJ pooled 23 trials with 1,357 adults with type 2 diabetes that compared low-carbohydrate diets, under 130 grams a day, and very low carbohydrate diets, under 50 grams a day, with control diets that were mostly low in fat. Of the 23 trials, 12 met the very low carbohydrate definition. At 6 months, HbA1c, a measure of average blood sugar, was 0.47 percentage points lower on the carbohydrate-restricted diets, from 0.34 to 0.60 lower, rated high in certainty using GRADE. At 12 months the difference was about half as large, 0.23 lower, from 0.46 lower to no difference, rated moderate (Goldenberg et al., 2021).

The same review found that more people reached an HbA1c below 6.5 percent at 6 months on the carbohydrate-restricted diets, about 59 per 100 against 31 per 100, rated moderate. When the definition also required being off all diabetes medicine, the difference was small and uncertain, 16 per 100 against 13 per 100, a risk ratio of 1.24, from 0.65 to 2.38, rated low; at 12 months on that stricter definition the estimate pointed slightly the other way. Remission was less common in trials that included people using insulin (Goldenberg et al., 2021). These are pooled results for both low and very low carbohydrate diets, not for ketogenic diets alone.

A 2022 review limited to very low carbohydrate or ketogenic diets, 50 grams a day or less, found 8 trials of at least 6 months with 606 adults with type 2 diabetes or prediabetes. The trials reported their results in ways that could not all be combined. Of these, 4 reported changes from the start and together did not show a clear difference in HbA1c at 12 months, 0.01 percentage points, from 0.22 lower to 0.25 higher; 2 others, reported as final values, showed a lower HbA1c, 0.65 percentage points lower, from 0.31 to 0.99 lower (Parry-Strong et al., 2022).

In the Keto-Med trial, 40 adults with prediabetes or type 2 diabetes began, and 33 completed, a comparison that had them eat a well-formulated ketogenic diet, 20 to 50 grams of carbohydrate a day, and a Mediterranean-style diet that also cut added sugar and refined grains but kept beans, fruit and whole grains, for 12 weeks each in random order. HbA1c fell on both diets, by 9 percent of its starting value on the ketogenic diet and 7 percent on the Mediterranean-style diet, and the trial did not demonstrate a difference between them (Gardner et al., 2022). Triglycerides fell more on the ketogenic diet, by 16 percent against 5 percent, and LDL cholesterol rose by 10 percent on the ketogenic diet while falling by 5 percent on the Mediterranean-style diet. Intake of fiber, thiamin, folate and magnesium was lower on the ketogenic diet (Gardner et al., 2022).

LDL cholesterol is the result that most separates the ketogenic diet from other healthy patterns. A 2026 review reported higher LDL cholesterol on ketogenic diets than on control diets, and rated that result low in certainty using GRADE. Its study table, however, includes a nonrandomized comparison with a baseline diet, diets allowing up to about 100 grams of carbohydrate a day, more than this page's ketogenic definition, and 2 reports that may describe the same trial. This page therefore does not use its pooled estimates as a ketogenic-only randomized comparison (Zhao et al., 2026).

In the 2021 diabetes review, LDL cholesterol at 12 months was 0.14 mmol/L higher on carbohydrate-restricted diets, a rise the authors judged clinically important although the interval reached no difference, rated moderate (Goldenberg et al., 2021).

The average hides wide variation. A 2026 review in the Journal of the American Heart Association reports that most trials show small or no rises in LDL cholesterol, but that some show rises of 22 to 70 mg/dL, and that these large rises occur mainly in adults of normal weight, while LDL cholesterol in people with more severe obesity may stay level or fall (DiMattia & Petersen, 2026). The Cochrane review of low-carbohydrate diets found probably little to no difference in LDL cholesterol at 1 to 2 years, but most of its trials did not restrict carbohydrate to ketogenic levels (Naude et al., 2022). No trial found for this page measured heart attacks, strokes or deaths from cardiovascular disease, and the Cochrane review lists cardiovascular death as not measured (Naude et al., 2022).

Cancer

The popular claim that a ketogenic diet starves cancer rests mostly on laboratory and animal studies. In people, the trials are small. In a randomized phase II trial in 32 evaluable patients with untreated metastatic pancreatic cancer, all receiving chemotherapy, those also assigned to a supervised ketogenic diet had a median overall survival of 13.7 months against 10.2 months, but the hazard ratio of 0.58, from 0.25 to 1.37, includes no difference, and the authors state that the trial was not powered for definitive outcomes (Jameson et al., 2026).

A 2025 review of controlled trials in people with cancer reported lower body fat, insulin and fatigue on ketogenic diets but did not report survival, and judged the methodological quality of most of its trials low (Zhang et al., 2025). For glioblastoma, a 2026 review of 41 studies, from small trials to case series, reported longer survival in patients who stayed in ketosis than in historical controls (Firdous et al., 2026); a comparison with past patients cannot separate the diet from differences in who was able to follow it. None of these studies establishes that a ketogenic diet prevents cancer or extends survival.

Body weight, as measured

This archive reports body weight as something a study measured, in that study's own numbers, and not as a goal or as a measure of whether a way of eating worked; see the standing rule in the review record at the foot of this page. The one exception is weight a person loses without meaning to, which can signal a health problem and is a reason to involve a clinician, never a sign that a diet worked.

The Cochrane review of low-carbohydrate diets pooled 61 trials with 6,925 adults with overweight or obesity, comparing low-carbohydrate with balanced-carbohydrate weight-reducing diets; in most trials both groups were asked to cut calories. Only 14 of the 61 trials restricted carbohydrate to 50 grams a day or less. In adults without type 2 diabetes, weight change at 1 to 2 years differed by 0.93 kilograms, from 1.81 to 0.04 kilograms, in favor of the low-carbohydrate diets, which the review rated as probably little to no difference, moderate in certainty; in adults with type 2 diabetes the difference was 0.33 kilograms, from 2.13 lower to 1.46 higher, also moderate (Naude et al., 2022).

In a subgroup of 9 trials with 494 participants that restricted carbohydrate to ketogenic levels, weight at 3 to under 12 months was 2.29 kilograms lower than on balanced diets, from 1.13 to 3.45 kilograms lower, against 0.36 kilograms lower in 24 trials of non-ketogenic low-carbohydrate diets. The review did not rate that subgroup. It judged that the subgroups did not show important clinical differences in average effects, which is a judgment about clinical size, not a finding that the subgroups were statistically alike (Naude et al., 2022).

An earlier review of 13 trials that lasted at least 12 months and initially prescribed very low carbohydrate diets, compared with low-fat diets, found weight 0.91 kilograms lower on the very low carbohydrate diets, from 0.17 to 1.65 kilograms lower, and LDL cholesterol 0.12 mmol/L higher (Bueno et al., 2013). In most of its trials, reported carbohydrate intake had risen above ketogenic levels by the end, so these results do not show the effect of staying in ketosis. In type 2 diabetes, the 2021 review found weight 3.46 kilograms lower at 6 months on carbohydrate-restricted diets and no clear difference at 12 months, 0.29 kilograms higher, from 1.02 lower to 1.60 higher, both rated moderate (Goldenberg et al., 2021).

The National Institutes of Health ran 2 tightly controlled studies of how the diet affects energy balance. When 17 men switched, in a fixed order rather than a random one, from a high-carbohydrate diet to a ketogenic diet with the same calories, they burned slightly more energy, near the limits of what could be detected, but did not lose body fat faster (Hall et al., 2016).

In a crossover trial, 21 adults were randomized to the order of 2 diets, each offered for 2 weeks with as much food as they wanted; one withdrew after low blood sugar during the ketogenic diet. Among the 20 who completed both periods, energy intake was about 689 calories a day lower on the plant-based, low-fat diet than on the animal-based ketogenic diet (Hall et al., 2021). Both were short inpatient studies designed to test a theory, not to compare diets for long-term weight, and published correspondence has since debated whether 2-week crossover periods like those of the 2021 study can give valid results (Hall, 2025; Ludwig et al., 2025).

Mortality

The searches for this page found no randomized trial of a ketogenic diet for weight, diabetes or general health that measured deaths, and the Cochrane review of low-carbohydrate diets lists cardiovascular death as not measured (Naude et al., 2022). In the epilepsy trials, deaths were rare and were not an outcome the trials were designed to estimate; the 3 deaths in the infant trial described above were judged unrelated to the diet (Schoeler et al., 2023).

Eating behavior and wellbeing

In the one epilepsy trial that measured it, no clear difference in quality of life was detected between children on the diet and children on usual care, and children on the diet were rated as more active, more productive and less anxious, at very low certainty (Martin-McGill et al., 2020). In type 2 diabetes, the 2021 review found no clear difference in diabetes-related quality of life at 6 months and a worsening at 12 months that the authors judged clinically important but that was not statistically clear, from 1 trial, rated low (Goldenberg et al., 2021). Early symptoms, often called keto flu, are described in section 6.

Evidence from long-term prospective cohort studies

A cohort study follows people who already eat a certain way and records what happens to them. It can show that a pattern goes with an outcome; it cannot by itself show that the pattern caused it. For ketogenic diets this kind of evidence is thin. Cohort studies of carbohydrate intake score people on how low their carbohydrate is relative to others, and even the lowest groups usually eat far more carbohydrate than a ketogenic diet allows, so their results are evidence about low-carbohydrate eating and apply to ketogenic diets only indirectly.

Cardiometabolic outcomes

In 3 large U.S. cohorts of health professionals followed for up to about 30 years, with 20,033 cases of coronary heart disease, people with the highest low-carbohydrate diet scores had slightly more heart disease than those with the lowest, a hazard ratio of 1.05, from 1.01 to 1.10. The direction depended on food quality: scores built on animal foods and refined carbohydrate went with more coronary heart disease, and a healthy low-carbohydrate score built on plant foods and whole grains went with less, a hazard ratio of 0.85, from 0.82 to 0.89 (Wu et al., 2026). The search for this page found no cohort that followed people on a ketogenic diet for heart attacks or strokes.

Cancer

The search for this page found no prospective cohort that followed people eating a ketogenic diet for cancer incidence, and the studies reviewed here do not establish an effect in either direction.

Body weight, as measured

The search for this page found no long-term cohort of people eating a ketogenic diet that reported body weight; the longest follow-up of a ketogenic program is the non-randomized study described below.

Mortality

In the U.S. Atherosclerosis Risk in Communities cohort, 15,428 adults followed for a median of 25 years showed a U-shaped pattern: the lowest risk of death was at 50 to 55 percent of energy from carbohydrate. In a combined analysis with other cohorts covering 432,179 people, carbohydrate intake below 40 percent of energy was associated with higher mortality than moderate intake, a hazard ratio of 1.20, from 1.09 to 1.32. Replacing carbohydrate with animal fat and protein went with higher mortality, and replacing it with plant fat and protein went with lower mortality (Seidelmann et al., 2018). In a national U.S. survey cohort of 37,233 adults, an overall low-carbohydrate score was not associated with mortality, while an unhealthy low-carbohydrate score was associated with higher mortality and a healthy one with lower (Shan et al., 2020). In all of these, "low" means well above ketogenic levels.

Evidence from non-randomized studies and case series

The most cited long-term evidence for the popular use comes from a study of adults with type 2 diabetes who were not assigned by chance to either a ketogenic program delivered through an app with remote clinicians or their usual care. Of 262 people in the program, 74 percent remained at 2 years; they had lost an average of 10 percent of body weight, and use of diabetes medicines other than metformin fell from 55.7 to 26.8 percent of participants, while LDL cholesterol rose. The average blood ketone level measured in the laboratory at 1 and 2 years was 0.27 mmol/L, below the 0.5 mmol/L the program set as its target (Athinarayanan et al., 2019).

At 5 years, 122 of the original participants remained, and 20 percent of them met the study's remission definition (McKenzie et al., 2024). The study had no randomized comparison, most of its authors were affiliated with the company that delivers the program, and the 5-year results come from the people who stayed.

At the other end of the scale are case series. A chart review of 17 cardiology patients whose LDL cholesterol rose above 190 mg/dL on a ketogenic diet found an average LDL of 316 mg/dL on the diet, ranging up to 810 mg/dL, against an average of 129 mg/dL before; 2 of the 5 who were tested carried a gene variant for familial hypercholesterolemia (Schmidt et al., 2023). The paper's percentage figures do not match its own mg/dL values, so this page reports only the latter. A cohort of 100 adults on long-term ketogenic diets with very high LDL cholesterol was studied with heart scans; according to a 2026 review, its primary publication was later retracted because of concerns about methods and data reliability, and the study had no comparison group (DiMattia & Petersen, 2026).

Highlighted studies

These 4 studies carry the most weight for this page: one review for the medical use, and a review, a trial and a second review for the popular use. Each is described so that a reader can judge it without opening the paper.

On a narrow screen, scroll the table sideways to see every column.

Highlighted studies for the Ketogenic Diet

Study	Design	Sample size	Duration	Finding	The limitation that matters
Martin-McGill et al., 2020, Cochrane review of ketogenic diets for drug-resistant epilepsy	Systematic review and meta-analysis of randomized trials	13 trials, 932 people: 711 children and 221 adults	Trials of 2 to 16 months; main results at 3 to 4 months in children	In children, a 50 percent or greater reduction in seizures in about 453 per 1,000 on a ketogenic diet against 78 per 1,000 on usual care (risk ratio 5.80, from 3.48 to 9.65), rated low in certainty. In adults on the modified Atkins diet the estimate was too imprecise to show a benefit, rated very low	No trial could be blinded, most were small and single-center, and the search ran to April 2019, before the infant trial reported in 2023
Goldenberg et al., 2021, BMJ review of low and very low carbohydrate diets in type 2 diabetes	Systematic review and meta-analysis of randomized trials, with unpublished data from trial authors	23 trials, 1,357 adults with type 2 diabetes	3 months to 2 years; results at 6 and 12 months	HbA1c 0.47 percentage points lower at 6 months (high certainty) and 0.23 lower at 12 months (moderate). Remission without diabetes medicine at 6 months uncertain, risk ratio 1.24, from 0.65 to 2.38 (low)	It pooled low-carbohydrate diets, under 130 grams a day, with very low carbohydrate diets, and most comparison diets were low in fat, so it does not isolate the ketogenic diet or compare it with other healthy patterns
Gardner et al., 2022, the Keto-Med trial	Randomized crossover trial	40 adults with prediabetes or type 2 diabetes began; 33 completed both phases	12 weeks on each diet, then 12 weeks of free choice	HbA1c fell on both diets, 9 percent and 7 percent of starting values, with no difference demonstrated between them. LDL cholesterol rose 10 percent on the ketogenic diet and fell 5 percent on the Mediterranean-style diet; triglycerides fell more on the ketogenic diet	Small and short, with meals supplied only for the first 4 weeks of each phase and several participants' phases extended by COVID-19 disruptions
Naude et al., 2022, Cochrane review of low-carbohydrate versus balanced-carbohydrate weight-reducing diets	Systematic review and meta-analysis of randomized trials	61 trials, 6,925 adults with overweight or obesity	3 months to 2 years	Probably little to no difference in weight at 1 to 2 years, about 1 kilogram or less, in adults with or without type 2 diabetes (moderate certainty). A subgroup of 9 ketogenic-level trials showed 2.29 kilograms more weight lost at 3 to under 12 months, not rated	Only 14 of the 61 trials restricted carbohydrate to ketogenic levels, and most results came from trials at high risk of bias, mostly from missing outcome data

Head to head

The ketogenic diet has been compared directly with several other patterns, and the comparison diet matters as much as the ketogenic one. Against a Mediterranean-style diet that also cut added sugar and refined grains, the Keto-Med trial did not demonstrate a difference in HbA1c after 12 weeks; the ketogenic diet lowered triglycerides more and raised LDL cholesterol, while the Mediterranean-style diet supplied more fiber and several vitamins and minerals and was the one participants leaned toward when free to choose (Gardner et al., 2022). The Mediterranean pattern has its own page in this archive: Mediterranean Diet.

Against low-fat and balanced diets, the comparisons are described above. In type 2 diabetes, trials pooling low and very low carbohydrate diets found an HbA1c difference that was smaller by 12 months, a weight difference that was no longer clear by then, and triglycerides that stayed lower; and in adults with overweight or obesity, low-carbohydrate diets as a group probably make little to no difference to weight at 1 to 2 years (Goldenberg et al., 2021; Naude et al., 2022). For epilepsy in infants, the comparison was a further antiseizure medicine, and the trial did not demonstrate a difference (Schoeler et al., 2023). For endurance athletes, the comparison was diets higher in carbohydrate, which supported better race performance (Burke et al., 2017).

Where the evidence is absent

The searches for this page found no randomized trial of a ketogenic diet followed for weight, diabetes or general health that measured heart attacks, strokes, new cancers or death, and no prospective cohort that followed people eating at ketogenic levels for those outcomes (the cancer-treatment trials described above are a separate use); the studies reviewed here do not establish an effect on them in either direction. Most popular-use trials lasted 3 to 12 months, a few ran to 2 years, and the longest follow-up, 5 years, comes from a study that was not randomized. Many trials described as ketogenic ended with participants eating more carbohydrate than the diet called for, so their results describe low-carbohydrate eating as much as ketosis.

For epilepsy, the trial evidence in adults is limited and imprecise, and long-term outcomes of years on the diet come from observational reports rather than trials. The searches found no trial of the diet in pregnancy, only case reports and a survey of clinicians; limited short-term randomized evidence in children and adolescents with obesity, without established long-term safety; and, in older adults, mainly research in people with cognitive impairment. Whether the large LDL cholesterol rises seen in some lean adults lead to heart disease over decades has not been answered by the studies reviewed here. The claims made for cancer treatment and brain health rest on small studies, and this page does not rate them.

5. Mechanisms

(optional reading; select the heading to open or close this section)

This section explains how the diet is thought to work in the body. You can skip it. Nothing in the coaching sections depends on it.

Established mechanisms

When carbohydrate intake is very low and insulin levels fall, the body releases more fat from storage, and the liver converts some of it into ketone bodies, chiefly beta-hydroxybutyrate and acetoacetate. Insulin holds this process back, and the hormones glucagon and epinephrine push it forward (Al Alawi et al., 2020). The ketones circulate in the blood and become a main source of energy for the brain when carbohydrate is scarce, as they do during fasting (Al Alawi et al., 2020). This is the state the diet is named for. In the clinical studies described on this page, programs set a target blood level of beta-hydroxybutyrate of 0.5 to 3.0 mmol/L for nutritional ketosis (Athinarayanan et al., 2019; Jameson et al., 2026).

Nutritional ketosis is not the same as ketoacidosis. Diabetic ketoacidosis is a medical emergency that requires high blood sugar or known diabetes, raised ketones and acid buildup in the blood together, and a beta-hydroxybutyrate level of 3.0 mmol/L or higher corresponds well with it (Umpierrez et al., 2024). In the studies described on this page, people on ketogenic diets had average levels far below that; in the 2-year diabetes program the average laboratory value was 0.27 mmol/L (Athinarayanan et al., 2019). The exceptions, which matter for coaching, are people taking SGLT2 inhibitors, in whom ketoacidosis can develop with near-normal blood sugar, and breastfeeding mothers who restrict carbohydrate or calories sharply (Al Alawi et al., 2020; Umpierrez et al., 2024).

Several effects on blood measures have been shown repeatedly in trials. Very low carbohydrate diets lower blood sugar and triglycerides and modestly raise HDL cholesterol, and they raise LDL cholesterol on average, with large individual differences (DiMattia & Petersen, 2026; Gardner et al., 2022; Goldenberg et al., 2021). Low-carbohydrate diets also increase the loss of water and salt in urine, especially in the early weeks, which is the reason for advice to keep up fluid and salt (Dyńka et al., 2026; Gardner et al., 2022; Kossoff et al., 2018).

On energy balance, a tightly controlled study at the National Institutes of Health did not find that a ketogenic diet causes more body fat to be lost when calories are held the same (Hall et al., 2016). Short inpatient studies like this one test a theory under controlled conditions; they do not show what happens over months of everyday eating.

Proposed mechanisms

How the diet reduces seizures is not settled. Researchers have proposed several mechanisms acting together, including effects on the brain's signaling chemicals, on how brain cells produce energy, on the acidity of brain tissue and on inflammation, and they note that ketones alone do not explain the effect (Auvin, 2021; Kossoff et al., 2018).

Why LDL cholesterol rises so steeply in some lean adults is also unsettled. Proposed explanations include greater sensitivity of the liver's cholesterol handling in people with less body fat and a model in which a greater flow of fat through the liver raises the output of cholesterol-carrying particles, called the lipid energy model. A 2026 review describes these as hypotheses (DiMattia & Petersen, 2026).

The cancer claim rests on the observation that many tumors rely heavily on glucose, so that lowering glucose and raising ketones might slow them; this idea comes from laboratory and animal research, and the human trials described in section 4 have not shown that it works in people (Firdous et al., 2026; Jameson et al., 2026). A mechanism shown in cells, in animals or in a change of a blood marker is a reason to study a diet, not evidence that the diet prevents or treats a disease.

6. Risks, contraindications and myths

(select the heading to open or close this section)

Warnings from professional or regulatory bodies

The American Diabetes Association's 2026 Standards of Care state that very low carbohydrate eating plans "should be avoided" by people taking SGLT2 inhibitors "because of the potential risk of ketoacidosis," and that such plans "are not currently recommended for individuals who are pregnant or lactating, children, people who have kidney disease, or people with or at risk for disordered eating." The same document asks clinicians to "maintain consistent medical oversight" of people following very low carbohydrate plans, because insulin and other diabetes medicines may need adjusting (American Diabetes Association Professional Practice Committee for Diabetes, 2026).

The American Heart Association's 2023 review of popular diets placed very low carbohydrate and ketogenic diets, with the paleo diet, in its bottom tier, describing them as "patterns of strong concern" that "promote restriction of food groups that are considered essential features of a heart-healthy diet (legumes, whole grains) and allow high-saturated-fat sources that are strongly discouraged" (Gardner et al., 2023). Its 2026 dietary guidance adds that "some popular weight loss diets may result in short-term benefits but have uncertain long-term cardiovascular impact and can worsen cardiovascular risk factors," citing a case series of dramatic LDL cholesterol rises on ketogenic diets (Lichtenstein et al., 2026).

Nutrient gaps

The diet removes or sharply limits fruit, whole grains, beans and starchy vegetables, so the nutrients concentrated in those foods are harder to get. In the Keto-Med trial, participants took in less fiber, thiamin, folate and magnesium on the ketogenic diet than on the Mediterranean-style diet (Gardner et al., 2022). The American Heart Association suggests that people following a ketogenic diet consider a multivitamin and mineral supplement including potassium and magnesium (Gardner et al., 2023). In the medical diet for epilepsy, supplementation is treated as essential: children receive a carbohydrate-free multivitamin with minerals, especially selenium, plus calcium and vitamin D (Kossoff et al., 2018).

Folate matters particularly before and during pregnancy. In a U.S. case-control study, carbohydrate intake in the year before conception was counted as restricted when it was in the lowest 5 percent of estimated intake among mothers of babies without birth defects, about 95 grams a day or less. Among those mothers, women with restricted intake took in less than half as much folic acid from food as the others. Women with restricted intake had slightly higher odds of a pregnancy affected by a neural tube defect, an adjusted odds ratio of 1.30, from 1.02 to 1.67; its authors point out that avoiding carbohydrate means eating fewer fortified foods (Desrosiers et al., 2018). Whether a particular plan meets a particular person's needs is a question for a registered dietitian.

Who should not follow it without medical supervision

For the medical diet, the international epilepsy consensus lists absolute contraindications: inherited disorders of carnitine and fat metabolism, pyruvate carboxylase deficiency and porphyria, in which the switch to burning fat can cause severe illness (Kossoff et al., 2018). The international adult recommendations add inherited high cholesterol, heart disease, low bone density, kidney stones, wasting and pregnancy as reasons for caution, and acute pancreatitis as a contraindication (Cervenka et al., 2021).

For the popular diet, the American Diabetes Association names people taking SGLT2 inhibitors, people who are pregnant or breastfeeding, children, people with kidney disease and people with or at risk for disordered eating, and asks for medical oversight of anyone on insulin or other diabetes medicines (American Diabetes Association Professional Practice Committee for Diabetes, 2026). A 2026 review adds advanced liver disease, inherited high cholesterol, gallbladder disease, heart rhythm problems, being underweight and recovery from surgery (Dyńka et al., 2026). Anyone in these groups needs a physician involved before starting.

Medication interactions

A coach never adjusts medication. The point of knowing these is to know when to send a client back to the prescriber.

Insulin and sulfonylureas lower blood sugar regardless of what is eaten, so a sudden drop in carbohydrate can cause low blood sugar unless doses are adjusted; the authors of the 2021 diabetes review make the same warning (American Diabetes Association Professional Practice Committee for Diabetes, 2026; Goldenberg et al., 2021). SGLT2 inhibitors can cause ketoacidosis when combined with very low carbohydrate eating, in published cases within days of starting the drug or the diet, and with near-normal blood sugar (Mistry & Eschler, 2021; Umpierrez et al., 2024). Blood pressure medicines may need review, and blood pressure needs monitoring (American Diabetes Association Professional Practice Committee for Diabetes, 2026).

In the medical diet, the epilepsy team manages interactions with antiseizure medicines: topiramate, zonisamide and acetazolamide can add to the mild acid buildup the diet causes, so bicarbonate is checked, and antiseizure medicines are often reduced once the diet is working, under the neurologist's direction (Kossoff et al., 2018). For warfarin, the search for this page found one report of 2 patients who needed higher warfarin doses after starting high-protein, low-carbohydrate diets, which its authors rated a possible interaction (Beatty et al., 2005); a client on warfarin keeps the prescriber informed before a large change in eating.

Signs it is going wrong

Physical signs a coach or a family member might notice include headache, tiredness, poor sleep, irritability, nausea and constipation in the first days, the cluster called keto flu, which usually eases (Gardner et al., 2023); a kidney stone the client reports, since a review of 36 studies estimated that about 6 in 100 people on ketogenic diets developed one over an average of about 4 years, although nearly all of those studies were of medical ketogenic diets, mostly in children, so the figure is not a direct estimate for an otherwise healthy adult on the popular diet (Acharya et al., 2021); changes in menstrual periods, which adult epilepsy centers report (Cervenka et al., 2021); and weight loss the person did not intend.

A client on an SGLT2 inhibitor who has vomiting, heavy thirst and urination, or confusion needs urgent care (Umpierrez et al., 2024), and so does a breastfeeding mother on a strict diet with nausea, vomiting or abdominal pain (Al Alawi et al., 2020). Each is a reason to involve a clinician, not a diagnosis; working out the cause is the clinician's job.

Behavior matters as much, and has to be read with care. Counting carbohydrate is ordinary on this diet, and for a medical diet it is the prescription, so on its own it tells a coach little. Rules that keep getting stricter, distress after eating a food outside the plan, or avoiding meals with others carry more meaning and are worth asking about directly: what the behavior is for, whether it causes distress, whether it is narrowing what the person eats, and whether it is getting in the way of daily life. Where the answers raise a real concern about disordered eating, the coach follows their own referral process.

Myths and where they came from

Myth: keto burns fat in a way other diets cannot

Where it came from: the idea that low insulin unlocks fat stores, called the carbohydrate-insulin model, and the rapid early weight change many people see. What is true: in a tightly controlled metabolic ward study, a ketogenic diet with the same calories as a high-carbohydrate diet did not increase body fat loss (Hall et al., 2016), and in a 2-week inpatient study in which people could eat freely, they ate more on a ketogenic diet than on a plant-based, low-fat one, a design whose validity has since been debated (Hall, 2025; Hall et al., 2021; Ludwig et al., 2025). Over 1 to 2 years, low-carbohydrate diets as a group probably make little to no difference to weight compared with balanced diets (Naude et al., 2022).

Myth: keto reverses type 2 diabetes

Where it came from: real improvements in blood sugar, programs that market diabetes "reversal," and personal stories of stopping medicine. What is true: carbohydrate-restricted diets lower HbA1c more than comparison diets over 6 months, at high certainty, but the difference is about halved by 12 months, and the evidence that more people reach normal blood sugar without any diabetes medicine is low in certainty (Goldenberg et al., 2021).

Myth: a high cholesterol number on keto does not matter

Where it came from: suggestions that the rise is harmless in people who are otherwise metabolically healthy, and the fact that triglycerides and HDL cholesterol often improve at the same time (DiMattia & Petersen, 2026). What is true: LDL cholesterol is a causal risk factor for cardiovascular disease, and the American Heart Association states that ketogenic diets are high in fat without limiting saturated fat, "leading to increases in low-density lipoprotein cholesterol" (Gardner et al., 2023; Lichtenstein et al., 2026).

Myth: ketosis is the same as ketoacidosis, or the reverse, that ketones can never be dangerous

Where it came from: the similar names, and the fact that both involve ketones. What is true: in the studies on this page, diet-induced ketosis kept average blood ketones far below the levels seen in diabetic ketoacidosis, a medical emergency that also involves acid buildup in the blood (Athinarayanan et al., 2019; Umpierrez et al., 2024). But ketoacidosis can occur with near-normal blood sugar in people taking SGLT2 inhibitors, and it has been reported in breastfeeding mothers on strict diets (Al Alawi et al., 2020; Umpierrez et al., 2024).

Myth: keto starves cancer

Where it came from: laboratory findings that many tumors depend heavily on glucose, and promotional summaries of small trials, such as a company headline that its nutrition therapy "extends survival by 34%" in advanced pancreatic cancer (Virta Health, n.d.). What is true: that trial had 32 evaluable patients and a hazard ratio for death of 0.58, from 0.25 to 1.37, an interval that includes no difference, and its authors state it was not powered for definitive outcomes (Jameson et al., 2026). No study found for this page establishes that a ketogenic diet prevents cancer or extends survival.

7. Key terms

Ketone bodies, or ketones

Substances the liver makes from fat when carbohydrate and insulin are low, chiefly beta-hydroxybutyrate and acetoacetate, which the body can use for energy.

Nutritional ketosis

The state of raised blood ketones produced by a ketogenic diet; programs in the studies on this page targeted 0.5 to 3.0 mmol/L of beta-hydroxybutyrate.

Ketoacidosis

A medical emergency in which ketones build up with acid in the blood; in diabetes it usually comes with high blood sugar, but on SGLT2 inhibitors it can occur with near-normal blood sugar.

Classic ketogenic diet

The original medical diet for epilepsy, built on a ratio of about 4 grams of fat to 1 gram of protein and carbohydrate combined, with foods weighed and the menu calculated by a dietitian.

Modified Atkins diet

A less restrictive medical ketogenic diet that limits carbohydrate, starting at about 10 to 15 grams a day, without weighing foods or fixing a ratio.

Drug-resistant epilepsy

Epilepsy in which seizures continue despite trials of antiseizure medicines; the United Kingdom trial in section 4 enrolled children who had not responded to at least 2.

Well-formulated ketogenic diet

A popular-use version that keeps carbohydrate at about 20 to 50 grams a day while emphasizing nonstarchy vegetables and quality fats.

HbA1c

A blood test that reflects average blood sugar over recent months; it is the main outcome in diabetes trials.

Remission of type 2 diabetes

Blood sugar back in the normal or near-normal range; definitions differ on whether diabetes medicine is still allowed, and the difference matters when reading a claim.

SGLT2 inhibitor

A class of diabetes medicines, including empagliflozin, dapagliflozin and canagliflozin, that lowers blood sugar by passing glucose into the urine and that can cause ketoacidosis when combined with very low carbohydrate eating.

LDL cholesterol

The cholesterol carried in low-density lipoprotein particles, a causal risk factor for cardiovascular disease, which rises on average on ketogenic diets.

Keto flu

A cluster of temporary symptoms, such as headache, fatigue, poor sleep, irritability, nausea and constipation, reported in the first days of a ketogenic diet.

Exogenous ketones

Supplements sold to raise blood ketones without changing the diet; the international epilepsy consensus finds no evidence for their role in treating epilepsy.

Subgroup analysis

A look at results within part of a review's trials, such as only the ketogenic ones; it can suggest a difference but is less reliable than the main comparison.

GRADE

The 4-level scale used in clinical guidelines to say how certain the evidence for a conclusion is: high, moderate, low or very low. It rates the evidence for one outcome and one comparison at a time, not the diet.

8. Review questions

1. In 3 or 4 sentences, explain how the medical ketogenic diet for epilepsy and the popular ketogenic diet differ in who prescribes them, how carbohydrate is set, and how they are monitored.
2. Using section 4, describe what the Cochrane review of epilepsy trials found for children compared with usual care: how many trials and children, which outcome, how large the difference was, how long the trials lasted, and why the certainty was rated low.
3. A client with type 2 diabetes tells you keto "reversed" her neighbor's diabetes. Using the 2021 review in section 4, explain the difference between the 2 definitions of remission it used and what it found for each at 6 months.
4. A client says, "I'm starting keto Monday. I take metformin and Jardiance." Write what you would say and ask, and say exactly where the referral line is and why. (Jardiance is a brand of empagliflozin, an SGLT2 inhibitor.)
5. Where did the belief that a ketogenic diet burns fat in a way other diets cannot come from, and what did the tightly controlled studies in section 4 find? Write the 2 sentences you would use if a client raised it.

9. References

Acharya, P., Acharya, C., Thongprayoon, C., Hansrivijit, P., Kanduri, S. R., Kovvuru, K., Medaura, J., Vaitla, P., Garcia Anton, D. F., Mekraksakit, P., Pattharanitima, P., Bathini, T., & Cheungpasitporn, W. (2021). Incidence and characteristics of kidney stones in patients on ketogenic diet: A systematic review and meta-analysis. Diseases, 9(2), Article 39. https://doi.org/10.3390/diseases9020039

Al Alawi, A. M., Al Flaiti, A., & Falhammar, H. (2020). Lactation ketoacidosis: A systematic review of case reports. Medicina, 56(6), Article 299. https://doi.org/10.3390/medicina56060299

American Diabetes Association Professional Practice Committee for Diabetes. (2026). 5. Facilitating positive health behaviors and well-being to improve health outcomes: Standards of Care in Diabetes-2026. Diabetes Care, 49(Suppl. 1), S89-S131. https://doi.org/10.2337/dc26-S005

Athinarayanan, S. J., Adams, R. N., Hallberg, S. J., McKenzie, A. L., Bhanpuri, N. H., Campbell, W. W., Volek, J. S., Phinney, S. D., & McCarter, J. P. (2019). Long-term effects of a novel continuous remote care intervention including nutritional ketosis for the management of type 2 diabetes: A 2-year non-randomized clinical trial. Frontiers in Endocrinology, 10, Article 348. https://doi.org/10.3389/fendo.2019.00348

Auvin, S. (2021). New developments for dietary treatment of epilepsy after a century of history for the ketogenic diet. Brain Communications, 3(4), Article fcab234. https://doi.org/10.1093/braincomms/fcab234

Bao, W., Bowers, K., Tobias, D. K., Olsen, S. F., Chavarro, J., Vaag, A., Kiely, M., & Zhang, C. (2014). Prepregnancy low-carbohydrate dietary pattern and risk of gestational diabetes mellitus: A prospective cohort study. The American Journal of Clinical Nutrition, 99(6), 1378-1384. https://doi.org/10.3945/ajcn.113.082966

Beatty, S. J., Mehta, B. H., & Rodis, J. L. (2005). Decreased warfarin effect after initiation of high-protein, low-carbohydrate diets. The Annals of Pharmacotherapy, 39(4), 744-747. https://doi.org/10.1345/aph.1E454

Bueno, N. B., de Melo, I. S., de Oliveira, S. L., & da Rocha Ataide, T. (2013). Very-low-carbohydrate ketogenic diet v. low-fat diet for long-term weight loss: A meta-analysis of randomised controlled trials. British Journal of Nutrition, 110(7), 1178-1187. https://doi.org/10.1017/S0007114513000548

Burke, L. M., Ross, M. L., Garvican-Lewis, L. A., Welvaert, M., Heikura, I. A., Forbes, S. G., Mirtschin, J. G., Cato, L. E., Strobel, N., Sharma, A. P., & Hawley, J. A. (2017). Low carbohydrate, high fat diet impairs exercise economy and negates the performance benefit from intensified training in elite race walkers. The Journal of Physiology, 595(9), 2785-2807. https://doi.org/10.1113/JP273230

Cervenka, M. C., Wood, S., Bagary, M., Balabanov, A., Bercovici, E., Brown, M. G., Devinsky, O., Di Lorenzo, C., Doherty, C. P., Felton, E., Healy, L. A., Klein, P., Kverneland, M., Lambrechts, D., Langer, J., Nathan, J., Munn, J., Nguyen, P., Phillips, M., . . . Zupec-Kania, B. (2021). International recommendations for the management of adults treated with ketogenic diet therapies. Neurology: Clinical Practice, 11(5), 385-397. https://doi.org/10.1212/CPJ.0000000000001007

de Oliveira, J., de Paula, A. C. P., Dos Santos, T. D. S. M., Smaira, F. I., Caruso, B. M., & Guimarães, V. H. D. (2025). Worsening binge eating over time following an unsupervised low-carbohydrate diet. Nutrition, 140, Article 112906. https://doi.org/10.1016/j.nut.2025.112906

Desrosiers, T. A., Siega-Riz, A. M., Mosley, B. S., & Meyer, R. E. (2018). Low carbohydrate diets may increase risk of neural tube defects. Birth Defects Research, 110(11), 901-909. https://doi.org/10.1002/bdr2.1198

DiMattia, Z. S., & Petersen, K. S. (2026). Ketogenic diets and low-density lipoprotein cholesterol in adults with normal weight: An emerging clinical challenge. Journal of the American Heart Association, 15(9), Article e048903. https://doi.org/10.1161/JAHA.125.048903

Dyńka, D., Rodzeń, Ł., Rodzeń, M., Łojko, D., Karakuła-Juchnowicz, H., Ede, G., Grzywacz, Ż., Antosik, K., Sethi, S., & Unwin, D. (2026). The ketogenic diet is not for everyone: Contraindications, side effects, and drug interactions. Annals of Medicine, 58(1), Article 2603016. https://doi.org/10.1080/07853890.2025.2603016

Faltersack, K., McDonald, T. J. W., De Giorgis, V., Barron, B., Cervenka, M. C., Devlin, A., Healy, L. A., Kverneland, M., van der Louw, E., Lu, M., Zanaboni, M. P., Armeno, M., & Felton, E. A. (2025). Ketogenic diet therapy for epilepsy during pregnancy and lactation: An international survey exploring clinician perspectives. Epilepsia Open, 10(4), 1120-1132. https://doi.org/10.1002/epi4.70067

Firdous, J., Asif, A. E., Haris, H. M., Iqbal, U., Khalid, M., Zahid, H. F., Mubarika, M., Abufatima, I. O., Amin, M. H. J., Hafeez, S., Jabeen, F., Shahzaib, M., Fatima, N., Kumar, N., Sen, S., & Hassan, A. (2026). Efficacy and safety of ketogenic diet in glioblastoma: An updated systematic review and meta-analysis. Neurological Sciences, 47(5), Article 461. https://doi.org/10.1007/s10072-026-09035-y

Frank, G. K. W., Scolnick, B., Beckwith, C., Shott, M. E., Kilty, S., Swindle, S., Lutter, M., & Rho, J. M. (2026). Symptom impact and safety of ketogenic therapy in adults with anorexia nervosa: A feasibility trial. Communications Medicine, 6(1), Article 315. https://doi.org/10.1038/s43856-026-01644-0

Gardner, C. D., Landry, M. J., Perelman, D., Petlura, C., Durand, L. R., Aronica, L., Crimarco, A., Cunanan, K. M., Chang, A., Dant, C. C., Robinson, J. L., & Kim, S. H. (2022). Effect of a ketogenic diet versus Mediterranean diet on glycated hemoglobin in individuals with prediabetes and type 2 diabetes mellitus: The interventional Keto-Med randomized crossover trial. The American Journal of Clinical Nutrition, 116(3), 640-652. https://doi.org/10.1093/ajcn/nqac154

Gardner, C. D., Vadiveloo, M. K., Petersen, K. S., Anderson, C. A. M., Springfield, S., Van Horn, L., Khera, A., Lamendola, C., Mayo, S. M., & Joseph, J. J. (2023). Popular dietary patterns: Alignment with American Heart Association 2021 dietary guidance: A scientific statement from the American Heart Association. Circulation, 147(22), 1715-1730. https://doi.org/10.1161/CIR.0000000000001146

Gasparre, N., Pasqualone, A., Mefleh, M., & Boukid, F. (2022). Nutritional quality of gluten-free bakery products labeled ketogenic and/or low-carb sold in the global market. Foods, 11(24), Article 4095. https://doi.org/10.3390/foods11244095

Goldenberg, J. Z., Day, A., Brinkworth, G. D., Sato, J., Yamada, S., Jönsson, T., Beardsley, J., Johnson, J. A., Thabane, L., & Johnston, B. C. (2021). Efficacy and safety of low and very low carbohydrate diets for type 2 diabetes remission: Systematic review and meta-analysis of published and unpublished randomized trial data. BMJ, 372, Article m4743. https://doi.org/10.1136/bmj.m4743

Hall, K. D. (2025). Reply to: Concern for the validity of short-term dietary crossover trials. Nature Medicine, 31(9), 2915-2916. https://doi.org/10.1038/s41591-025-03911-4

Hall, K. D., Chen, K. Y., Guo, J., Lam, Y. Y., Leibel, R. L., Mayer, L. E., Reitman, M. L., Rosenbaum, M., Smith, S. R., Walsh, B. T., & Ravussin, E. (2016). Energy expenditure and body composition changes after an isocaloric ketogenic diet in overweight and obese men. The American Journal of Clinical Nutrition, 104(2), 324-333. https://doi.org/10.3945/ajcn.116.133561

Hall, K. D., Guo, J., Courville, A. B., Boring, J., Brychta, R., Chen, K. Y., Darcey, V., Forde, C. G., Gharib, A. M., Gallagher, I., Howard, R., Joseph, P. V., Milley, L., Ouwerkerk, R., Raisinger, K., Rozga, I., Schick, A., Stagliano, M., Torres, S., . . . Chung, S. T. (2021). Effect of a plant-based, low-fat diet versus an animal-based, ketogenic diet on ad libitum energy intake. Nature Medicine, 27(2), 344-353. https://doi.org/10.1038/s41591-020-01209-1

Jameson, G. S., Roe, D. J., Borazanci, E., Hanna, D. L., Roberts, C. G. P., Pelster, M. S., Frank, R. C., Alistar, A. T., Miller, A. M., Wiedmeier-Nutor, J. E., Algaze, S. D., Zoller, A. R., Hallberg, S. J., Wertheim, B. C., Lee, K., Cridebring, D., Rabinowitz, J. D., Gately, S., Keppler, J., . . . Rasco, D. W. (2026). A randomized phase II trial of gemcitabine, nab-paclitaxel, cisplatin with or without a medically supervised ketogenic diet for patients with metastatic pancreatic cancer. Cancer, 132(6), Article e70343. https://doi.org/10.1002/cncr.70343

King, L., Maccarrone, J., Kaur, A., Chen, S., Kwan, P., O'Brien, T. J., & Kaul, N. (2026). A price check on the ketogenic diet for adults with epilepsy in Australia. Epilepsia Open, 11(4), 1237-1243. https://doi.org/10.1002/epi4.70285

Kirk, S., Brehm, B., Saelens, B. E., Woo, J. G., Kissel, E., D'Alessio, D., Bolling, C., & Daniels, S. R. (2012). Role of carbohydrate modification in weight management among obese children: A randomized clinical trial. The Journal of Pediatrics, 161(2), 320-327.e1. https://doi.org/10.1016/j.jpeds.2012.01.041

Kossoff, E. H., Zupec-Kania, B. A., Auvin, S., Ballaban-Gil, K. R., Bergqvist, A. G. C., Blackford, R., Buchhalter, J. R., Caraballo, R. H., Cross, J. H., Dahlin, M. G., Donner, E. J., Guzel, O., Jehle, R. S., Klepper, J., Kang, H. C., Lambrechts, D. A., Liu, Y. M. C., Nathan, J. K., Nordli, D. R., . . . Wirrell, E. C. (2018). Optimal clinical management of children receiving dietary therapies for epilepsy: Updated recommendations of the International Ketogenic Diet Study Group. Epilepsia Open, 3(2), 175-192. https://doi.org/10.1002/epi4.12225

Krebs, N. F., Gao, D., Gralla, J., Collins, J. S., & Johnson, S. L. (2010). Efficacy and safety of a high protein, low carbohydrate diet for weight loss in severely obese adolescents. The Journal of Pediatrics, 157(2), 252-258. https://doi.org/10.1016/j.jpeds.2010.02.010

Kovinthapillai, R., Lan, Y. Y., Kędzia, A., & Niechciał, E. (2026). Ketogenic diet in children with type 1 diabetes: Parental motivations and potential risks for metabolic health and development. Nutrients, 18(8), Article 1244. https://doi.org/10.3390/nu18081244

Li, S., Du, Y., Meireles, C., Sharma, K., Qi, L., Castillo, A., & Wang, J. (2023). Adherence to ketogenic diet in lifestyle interventions in adults with overweight or obesity and type 2 diabetes: A scoping review. Nutrition & Diabetes, 13(1), Article 16. https://doi.org/10.1038/s41387-023-00246-2

Lichtenstein, A. H., Khera, A., Anderson, C. A. M., Appel, L. J., DeSilva, D. M., Gardner, C., Hu, F. B., Jones, D. W., Petersen, K. S., & American Heart Association. (2026). 2026 dietary guidance to improve cardiovascular health: A scientific statement from the American Heart Association. Circulation, 153(18), e1285-e1295. https://doi.org/10.1161/CIR.0000000000001435

Lister, N. B., Melville, H., & Jebeile, H. (2024). What adolescents see on Instagram: Content analysis of #intermittentfasting, #keto, and #lowcarb. Nutrition & Dietetics, 81(3), 316-324. https://doi.org/10.1111/1747-0080.12853

Ludwig, D. S., Willett, W. C., & Putt, M. E. (2025). Concern for the validity of short-term dietary crossover trials. Nature Medicine, 31(9), 2913-2914. https://doi.org/10.1038/s41591-025-03909-y

Martin-McGill, K. J., Bresnahan, R., Levy, R. G., & Cooper, P. N. (2020). Ketogenic diets for drug-resistant epilepsy. Cochrane Database of Systematic Reviews, 2020(6), Article CD001903. https://doi.org/10.1002/14651858.CD001903.pub5

McKenzie, A. L., Athinarayanan, S. J., Van Tieghem, M. R., Volk, B. M., Roberts, C. G. P., Adams, R. N., Volek, J. S., Phinney, S. D., & Hallberg, S. J. (2024). 5-Year effects of a novel continuous remote care model with carbohydrate-restricted nutrition therapy including nutritional ketosis in type 2 diabetes: An extension study. Diabetes Research and Clinical Practice, 217, Article 111898. https://doi.org/10.1016/j.diabres.2024.111898

Mistry, S., & Eschler, D. C. (2021). Euglycemic diabetic ketoacidosis caused by SGLT2 inhibitors and a ketogenic diet: A case series and review of literature. AACE Clinical Case Reports, 7(1), 17-19. https://doi.org/10.1016/j.aace.2020.11.009

Naude, C. E., Brand, A., Schoonees, A., Nguyen, K. A., Chaplin, M., & Volmink, J. (2022). Low-carbohydrate versus balanced-carbohydrate diets for reducing weight and cardiovascular risk. Cochrane Database of Systematic Reviews, 2022(1), Article CD013334. https://doi.org/10.1002/14651858.CD013334.pub2

Neal, E. G. (2008). A randomized controlled trial of two ketogenic diets in the treatment of childhood epilepsy [Doctoral dissertation, University College London]. UCL Discovery. https://discovery.ucl.ac.uk/1444255/1/U591557.pdf

Neal, E. G., Chaffe, H., Schwartz, R. H., Lawson, M. S., Edwards, N., Fitzsimmons, G., Whitney, A., & Cross, J. H. (2008). The ketogenic diet for the treatment of childhood epilepsy: A randomised controlled trial. The Lancet Neurology, 7(6), 500-506. https://doi.org/10.1016/S1474-4422(08)70092-9

O'Malley, K., Willits-Smith, A., & Rose, D. (2023). Popular diets as selected by adults in the United States show wide variation in carbon footprints and diet quality. The American Journal of Clinical Nutrition, 117(4), 701-708. https://doi.org/10.1016/j.ajcnut.2023.01.009

Parry-Strong, A., Wright-McNaughton, M., Weatherall, M., Hall, R. M., Coppell, K. J., Barthow, C., & Krebs, J. D. (2022). Very low carbohydrate (ketogenic) diets in type 2 diabetes: A systematic review and meta-analysis of randomized controlled trials. Diabetes, Obesity and Metabolism, 24(12), 2431-2442. https://doi.org/10.1111/dom.14837

Paskaleva, I. N., Kaleva, N. N., Dimcheva, T. D., Markova, P. P., & Ivanov, I. S. (2025). Low-carbohydrate (ketogenic) diet in children with obesity: Part 1. Diet impact on anthropometric indicators and indicators of metabolic syndrome and insulin resistance. Diseases, 13(4), Article 94. https://doi.org/10.3390/diseases13040094

Schmidt, T., Harmon, D. M., Kludtke, E., Mickow, A., Simha, V., & Kopecky, S. (2023). Dramatic elevation of LDL cholesterol from ketogenic-dieting: A case series. American Journal of Preventive Cardiology, 14, Article 100495. https://doi.org/10.1016/j.ajpc.2023.100495

Schoeler, N. E., Marston, L., Lyons, L., Halsall, S., Jain, R., Titre-Johnson, S., Balogun, M., Heales, S. J. R., Eaton, S., Orford, M., Neal, E., Reilly, C., Eltze, C., Stephen, E., Mallick, A. A., O'Callaghan, F., Agrawal, S., Parker, A., Kirkpatrick, M., . . . Cross, J. H. (2023). Classic ketogenic diet versus further antiseizure medicine in infants with drug-resistant epilepsy (KIWE): A UK, multicentre, open-label, randomised clinical trial. The Lancet Neurology, 22(12), 1113-1124. https://doi.org/10.1016/S1474-4422(23)00370-8

Seidelmann, S. B., Claggett, B., Cheng, S., Henglin, M., Shah, A., Steffen, L. M., Folsom, A. R., Rimm, E. B., Willett, W. C., & Solomon, S. D. (2018). Dietary carbohydrate intake and mortality: A prospective cohort study and meta-analysis. The Lancet Public Health, 3(9), e419-e428. https://doi.org/10.1016/S2468-2667(18)30135-X

Shan, Z., Guo, Y., Hu, F. B., Liu, L., & Qi, Q. (2020). Association of low-carbohydrate and low-fat diets with mortality among US adults. JAMA Internal Medicine, 180(4), 513-523. https://doi.org/10.1001/jamainternmed.2019.6980

Sondike, S. B., Copperman, N., & Jacobson, M. S. (2003). Effects of a low-carbohydrate diet on weight loss and cardiovascular risk factors in overweight adolescents. The Journal of Pediatrics, 142(3), 253-258. https://doi.org/10.1067/mpd.2003.4

Umpierrez, G. E., Davis, G. M., ElSayed, N. A., Fadini, G. P., Galindo, R. J., Hirsch, I. B., Klonoff, D. C., McCoy, R. G., Misra, S., Gabbay, R. A., Bannuru, R. R., & Dhatariya, K. K. (2024). Hyperglycemic crises in adults with diabetes: A consensus report. Diabetes Care, 47(8), 1257-1275. https://doi.org/10.2337/dci24-0032

van der Louw, E. J., Williams, T. J., Henry-Barron, B. J., Olieman, J. F., Duvekot, J. J., Vermeulen, M. J., Bannink, N., Williams, M., Neuteboom, R. F., Kossoff, E. H., Catsman-Berrevoets, C. E., & Cervenka, M. C. (2017). Ketogenic diet therapy for epilepsy during pregnancy: A case series. Seizure, 45, 198-201. https://doi.org/10.1016/j.seizure.2016.12.019

Virta Health. (n.d.). Virtual sustainable weight loss and diabetes reversal. Retrieved October 3, 2026, from https://www.virtahealth.com/

Wang, Y., Xiao, Q., Zhang, Z., & Yang, Y. (2025). Effects of ketogenic diet on muscle mass, strength, aerobic metabolic capacity, and endurance in adults: A systematic review and meta-analysis. Journal of Health, Population and Nutrition, 44(1), Article 346. https://doi.org/10.1186/s41043-025-01090-z

Wheless, J. W. (2008). History of the ketogenic diet. Epilepsia, 49(Suppl. 8), 3-5. https://doi.org/10.1111/j.1528-1167.2008.01821.x

Wu, Z., Liu, B., Wang, X., Alessa, H., Zeleznik, O. A., Eliassen, A. H., Clish, C., Wang, M., Mukamal, K. J., Rimm, E. B., Hu, Y., Hu, F. B., & Sun, Q. (2026). Effect of low-carbohydrate and low-fat diets on metabolomic indices and coronary heart disease in U.S. individuals. Journal of the American College of Cardiology, 87(20), 2764-2781. https://doi.org/10.1016/j.jacc.2025.12.038

Xi, L., Wan, J., Hua, Y., Guo, M., Du, J., Luo, X., & Yu, H. (2026). Effects of ketogenic diet on cognitive impairment in older adults: A systematic review and meta-analysis. Archives of Gerontology and Geriatrics, 151, Article 106409. https://doi.org/10.1016/j.archger.2026.106409

Zhang, M., Zhang, Q., Huang, S., Lu, Y., & Peng, M. (2025). Impact of ketogenic diets on cancer patient outcomes: A systematic review and meta-analysis. Frontiers in Nutrition, 12, Article 1535921. https://doi.org/10.3389/fnut.2025.1535921

Zhao, J., Wei, C., Gong, F., Pang, Z., & Zhao, H. (2026). Ketogenic diet-induced changes in adult lipid metabolism: A comprehensive systematic review and meta-regression of randomized controlled trials. BMC Cardiovascular Disorders, 26(1), Article 520. https://doi.org/10.1186/s12872-026-05799-5

Evidence rating and review record

Each row identifies the method, source, outcome or recommendation, and comparison being rated. Ratings from different tools are not interchangeable. The source is either a named published review or this page's own judgment. The rating in the banner is for this diet's main claimed outcome or recommendation, and the banner names the tool that produced it. A high rating means the research is solid; it does not mean the diet is right for everyone, and it does not mean a coach should recommend it.

Every rated row on this page was rated with GRADE, which uses the same 4-level scale as the clinical guidelines you will meet in practice. High means further research is very unlikely to change the conclusion. Moderate means further research may change it. Low means further research is likely to change it. Very low means the conclusion is uncertain. GRADE rates each outcome separately, so a diet can be high for one outcome and very low for another.

Three things about this table matter before reading it. The rows are split into the medical use and the popular use, and a rating in one part says nothing about the other. Several popular-use rows come from reviews that pooled low-carbohydrate diets with ketogenic ones; those rows say so, and their labels do not transfer to ketogenic diets alone. And the banner carries 2 ratings, one for each use, because no single rating can describe both.

Ratings marked as this page's own judgment were assigned from the evidence given in each row's reasons; ratings from a published review are reported exactly as that review published them.

On a narrow screen, scroll the table sideways to see every column.

Evidence ratings, their sources and reasons

Outcome	Certainty	Who rated it, and for which comparison	Why it is rated this way
Medical use. A 50 percent or greater reduction in seizures, children with drug-resistant epilepsy	Low	Martin-McGill et al., 2020, Cochrane review, using GRADE. Ketogenic diets (classic, classic with medium-chain triglyceride, medium-chain triglyceride, modified Atkins and simplified modified Atkins) against usual care, children aged 1 to 18, 4 trials, 385 children, 3 to 4 months	Randomized trials start high. Lowered 1 level for risk of bias, because some trials were not blinded, had missing data or reported their methods unclearly, and 1 level for imprecision, because the sample was small with fewer than 200 events. Risk ratio 5.80, from 3.48 to 9.65; about 453 against 78 per 1,000.
Medical use. Freedom from seizures, same children and comparison	Very low	Martin-McGill et al., 2020, Cochrane review, using GRADE. Same comparison, 4 trials, 385 children, 3 to 4 months	Lowered 1 level for risk of bias and 2 levels for imprecision, because there were fewer than 50 events. Risk ratio 3.16, from 1.20 to 8.35; about 66 against 21 per 1,000.
Medical use. A 50 percent or greater reduction in seizures, adults with drug-resistant epilepsy	Very low	Martin-McGill et al., 2020, Cochrane review, using GRADE. Modified Atkins diet against usual care, adults aged 16 and over, 2 trials, 141 adults, 2 to 3 months	Lowered for risk of bias, 2 levels for imprecision and 1 level for inconsistency between the 2 trials. Risk ratio 5.03, from 0.26 to 97.68, an interval that includes no benefit. No adult in either group became seizure-free.
Popular use. Body weight at 1 to 2 years, adults with overweight or obesity without type 2 diabetes, as measured	Moderate, for probably little to no difference	Naude et al., 2022, Cochrane review, using GRADE. Low-carbohydrate against balanced-carbohydrate weight-reducing diets, 14 trials, 1,805 adults. Most of the review's trials allowed more than 50 grams of carbohydrate a day, above ketogenic levels	Randomized trials start high. Lowered 1 level for risk of bias, because most of the information came from trials at high overall risk of bias. Mean difference 0.93 kilograms lower, from 1.81 to 0.04 kilograms lower, which the review judged little to no difference. This rating is for low-carbohydrate diets as a group and does not transfer to ketogenic diets alone. Reported as a measurement, not as a goal.
Popular use. Body weight at 1 to 2 years, adults with overweight or obesity and type 2 diabetes, as measured	Moderate, for probably little to no difference	Naude et al., 2022, Cochrane review, using GRADE. Same comparison, 7 trials, 813 adults	Lowered 1 level for risk of bias. Mean difference 0.33 kilograms lower, from 2.13 lower to 1.46 higher. For low-carbohydrate diets as a group, not ketogenic diets alone. Reported as a measurement, not as a goal.
Popular use. Body weight at 3 to under 12 months, ketogenic-level diets only, adults without type 2 diabetes, as measured	Low	This page's own judgment, using GRADE. The ketogenic subgroup of Naude et al., 2022: very low carbohydrate or ketogenic diets against balanced-carbohydrate weight-reducing diets, 9 trials, 494 adults. The review did not rate this subgroup	Randomized trials start high. Lowered 1 level for risk of bias, because 6 of the 9 trials were at high overall risk of bias, mostly from missing outcome data, and 3 had some concerns; the review lowered its main weight comparisons 1 level for the same kind of problem. Lowered 1 level for inconsistency: the ketogenic subgroup had moderate statistical heterogeneity, an I² of 51 percent, and materially different individual effects, from 4.60 kilograms lower to 2.20 kilograms higher, in the review's printed forest plot. Not lowered for indirectness: the comparison is the one named, ketogenic-level prescriptions against balanced diets. Not lowered for imprecision, judged against whether any mean weight difference exists: the interval, 1.13 to 3.45 kilograms lower, lies on one side of no difference. Publication bias was not detected but cannot be ruled out with 9 trials; not lowered. Reported as a measurement, not as a goal.
Popular use. HbA1c at 6 months, adults with type 2 diabetes	High	Goldenberg et al., 2021, using GRADE. Low and very low carbohydrate diets (under 130 grams a day) against control diets, mostly low-fat, 17 trials, 747 adults	Randomized trials start high and the review found no reason to lower the rating. Mean difference 0.47 percentage points lower, from 0.34 to 0.60 lower. The comparison pools low-carbohydrate and ketogenic diets, so it does not transfer to ketogenic diets alone.
Popular use. HbA1c at 12 months, same adults and comparison	Moderate	Goldenberg et al., 2021, using GRADE. Same comparison, 8 trials, 489 adults	Lowered 1 level for imprecision, because the interval includes a small effect, no effect and a small worsening. Mean difference 0.23 percentage points lower, from 0.46 lower to no difference.
Popular use. Diabetes remission, HbA1c below 6.5 percent with no diabetes medicine, at 6 months	Low	Goldenberg et al., 2021, using GRADE. Same comparison, 5 trials, 199 adults	Lowered for imprecision, because the optimal information size was not met and the interval is very wide. Risk ratio 1.24, from 0.65 to 2.38; about 16 against 13 per 100.
Popular use. HbA1c at 12 months, ketogenic-level diets only, adults with prediabetes or type 2 diabetes	Very low	This page's own judgment, using GRADE. Parry-Strong et al., 2022: very low carbohydrate or ketogenic diets, 50 grams a day or less, against higher-carbohydrate control diets; 6 trials reported HbA1c at 12 months. The review did not rate certainty	Randomized trials start high. Lowered 1 level for risk of bias, because outcome data were missing for many participants, with dropout from the ketogenic arms reaching 54 percent. Lowered 1 level for inconsistency, because the 2 sets of trials that had to be analyzed separately disagree, 0.01 percentage points higher in 4 trials and 0.65 lower in 2, although the trials within each set agreed. Lowered 1 level for indirectness, because reported carbohydrate intake in several trials rose above ketogenic levels, so the trials did not cleanly test a sustained ketogenic diet; that drift is counted here and not again under risk of bias. Imprecision is also serious: the set of 2 trials had only 66 participants, and its interval, 0.31 to 0.99 lower, includes values on both sides of a 0.5 percentage point reference difference, while the set of 4 trials, with 267 participants, sits close to no difference, from 0.22 lower to 0.25 higher. Publication bias could not be assessed with so few trials, and no extra level was taken for that alone. The rating is already at the lowest level, so the imprecision concern cannot lower it further; it is recorded so the appraisal is complete. The 0.5 reference is an aid to reading the numbers, not evidence that either diet changes health outcomes.
Popular use. LDL cholesterol at 12 months, adults with type 2 diabetes	Moderate	Goldenberg et al., 2021, using GRADE. Low and very low carbohydrate diets against control diets, 6 trials, 429 adults	Rated moderate as published. The review's supplement repeats the label and gives the 0.10 mmol/L threshold, but neither the supplement nor the main article reports which domain was lowered; the reason was not reported in the materials examined. Mean difference 0.14 mmol/L higher, from no difference to 0.28 higher, which the authors judged a clinically important harm beyond their threshold of 0.10 mmol/L, although the interval reaches no difference. Not a rating of ketogenic diets alone.
Popular use. Heart attacks, strokes and deaths	Not rated	This page's own judgment. Ketogenic diets followed for weight, diabetes or general health	No randomized trial found by the searches for this page measured these outcomes, and the Cochrane review of low-carbohydrate diets lists cardiovascular death as not measured. Cohort studies of low-carbohydrate scores studied carbohydrate intakes far above ketogenic levels, so they cannot supply a rating for the ketogenic diet. This is an absence of evidence, not a finding of no effect.

How this archive talks about weight. This page reports what studies measured, including body weight, and does not treat weight change as a goal or as a measure of a diet's success. Weight a person loses without meaning to can be a sign of a health problem, and where it is, the page names it as a reason to see a clinician. People are described the way the study described them.

Status: reviewed and approved by James Peth, MS, MPH, PhD, on October 4, 2026, after a combined final review of 5 archive pages, a focused recheck and a final confirmation. Written October 3, 2026, revised the same day after the combined final review, and repaired after the focused recheck that followed. The ratings marked as this page's own were assigned from the evidence.

What changed: After the combined final review: the evidence on children and adolescents now reports the short randomized trials of diets prescribed at ketogenic carbohydrate levels; the summaries of the diabetes trials now say that triglycerides stayed lower at 12 months and that those trials pooled low and very low carbohydrate diets.

In the same revision, the 2026 lipid review is no longer used for pooled numbers or as a rating row, because its study table mixes in a nonrandomized comparison, diets above ketogenic levels and possibly overlapping reports; the older-adult, athlete and long-term weight paragraphs now state what each review tested; the reasons for 2 of this page's own ratings were rewritten with every GRADE domain judged, without changing either label; the infant trial, the pregnancy case series and the kidney stone review were checked against their full texts; statements resting on the 2021 energy intake study were narrowed; the banner names the rating tool, outcome, comparison and source; and the mentions of related archive pages were revised. First version written October 3, 2026.

After the focused recheck: the 2003 adolescent trial now gives 39 randomized and 30 in the final analysis; the 2021 energy intake study now separates the 21 randomized from the 20 who completed both diets; the folate study is described as estimated intake with an adjusted odds ratio; the inconsistency reason for this page's own rating of weight in the ketogenic subgroup now rests on the review's printed forest plot, with the label unchanged; the introduction and caption of this review record follow the archive's current template; and the journal article on the United Kingdom epilepsy trial and the 2026 review in older adults, which could not be obtained, are marked as such where they are used.

Updated October 21, 2026: The American Heart Association is now named as an author of its 2026 dietary guidance statement (Lichtenstein et al., 2026), as the publisher's citation credits it. No content changed.

Next scheduled review: August 2027. Pages built on annual guidance are rechecked each year. The American Diabetes Association updates its Standards of Care every year, and its statements on very low carbohydrate eating and SGLT2 inhibitors should be rechecked first.

---

Diet Reference Archive | One James

Prepared by James Peth, MS, MPH, PhD

© 2026–2027 One James. All rights reserved.

This page is for education and coaching practice. It is not medical or dietary advice.