Low-Carb Diet for Diabetes: Remission, Safety & Who Funds It

Table of contents

• Evidence summary
• What counts as a low-carb diet, and what "remission" means
• How it works
• What works and what does not
• Benefits by claim
• Risks and all side effects
• All interactions with diabetes and other medicines
• Who should avoid a very-low-carb diet
• Carbohydrate levels studied, and keeping it up
• Follow the money: who sells it and who funded the evidence
• Related research
• Frequently asked questions
• Sources and funding notes

Evidence summary

Independent evidence and credibility scorecard

Independence tiers: 1 = regulator, government body or publicly funded review with no commercial stake; 2 = academic, charity or professional-society work with indirect ties (donated products, corporate partners, authors with industry links); 3 = interested party, such as a panel whose members have extensive industry ties; 4 = funded or written by the seller. Credibility runs from A (highest) to D.

Six of the twelve sources are US-based, three are UK-based and two more are partly UK, so this evidence reflects mainly a US and UK view. The opposing-interest signal is useful. SACN and Cochrane, which have no product to sell, agree with the low-carb-friendly BMJ meta-analysis on the main finding: low-carb works well for blood sugar in the first six months, and the advantage mostly fades by a year. Sources with different incentives point the same way, which raises our confidence.

What counts as a low-carb diet, and what "remission" means

There is no single agreed definition of a "low-carb" diet. SACN therefore compared "lower" with "higher" carbohydrate diets, not one fixed number (SACN). Research groups use different cut-offs:

• Low-carbohydrate diet: under 130 g a day, or under 26% of energy on a 2,000 kcal diet (Goldenberg 2021). The Cochrane review set the range more broadly, at more than 50 g up to 150 g a day, or under 45% of energy (Naude 2022).
• Very-low-carbohydrate or ketogenic diet: under 10% of energy from carbohydrate (Goldenberg), or 50 g a day or less (Cochrane). At this level the body makes ketones as an alternative fuel.
• Low-calorie diet (not the same thing): DiRECT used an 825–853 kcal-a-day formula of soups and shakes for 12 to 20 weeks, then reintroduced food in stages (Lean 2024). This is total diet replacement for weight loss, not carbohydrate restriction. People often confuse the two.

The 2021 definition of remission

An expert group convened by the ADA, including the European Association for the Study of Diabetes, Diabetes UK and the Endocrine Society, proposed "remission" as the preferred term. They defined it as HbA1c below 6.5% (48 mmol/mol) measured at least 3 months after stopping glucose-lowering medicines. After a lifestyle change, HbA1c should be tested at least 6 months after starting and 3 months after stopping medicines (Riddle 2021). The report avoids words such as "reversal" or "cure".

The definition matters when reading the evidence. Many low-carb trials count people as "in remission" while they still take medicines. Virta's own 5-year report includes a category of HbA1c under 6.5% "on no glucose lowering medication or only on metformin" (McKenzie 2024). That would not count as remission under the consensus definition.

How it works

Carbohydrate is the nutrient that raises blood glucose most directly after meals, so eating less of it lowers glucose readings. This explains the fast fall in HbA1c and fasting glucose in the first months. Low-carb diets lowered fasting glucose by 0.73 mmol/L more than control diets at six months (95% CI −1.19 to −0.27; n=611) (Goldenberg 2021). Triglycerides (blood fats made from excess carbohydrate) also fall. SACN found "adequate evidence" of a greater fall in triglycerides in the short term (SACN).

Remission works through a different route: weight loss. In DiRECT, remission rose steeply with weight lost. None of 76 people who gained weight went into remission, against 86% of the 36 people who lost 15 kg or more (Lean 2018). Why weight loss restores glucose control is still being researched, and we did not review the imaging studies behind the leading theory. Any diet that achieves enough lasting weight loss, low-carb or not, would be expected to work this way. When calories are matched, low-carb and balanced diets produce similar weight loss. In people with type 2 diabetes, the Cochrane review found only 1.26 kg more weight lost on low-carb diets at 3 to 6 months (95% CI −2.44 to −0.09; 14 trials, n=1,114), and little or no difference at 1 to 2 years (Naude 2022).

What is proven in humans is changes in blood tests: HbA1c, glucose, triglycerides, LDL and medicine use. What is not proven is any effect on heart attacks, strokes, kidney or eye damage, or death. SACN noted that "no trials" of low-carb diets in type 2 diabetes measured these outcomes (SACN).

What works and what does not

Benefits by claim

Does a low-carb diet lower blood sugar in type 2 diabetes?

Yes, in the short term. This is the most consistent finding in the field. The BMJ meta-analysis pooled 17 trials (n=747) and found HbA1c fell 0.47 percentage points more on low-carb diets than on control diets, mostly low-fat diets, at six months. GRADE rated this "high" certainty (Goldenberg 2021). That is roughly 5 mmol/mol. SACN, which has no commercial stake, reached the same verdict: "adequate evidence" of a greater HbA1c reduction over 3 to 6 months (SACN). The ADA now recommends considering lower carbohydrate intake for some adults with diabetes to improve glucose, and suggests limiting processed foods as one way to do it (grade B) (ADA 2026, recommendation 5.15).

Does the benefit last?

Less clearly. At 12 months the HbA1c advantage roughly halved to −0.23%, and its confidence interval touched zero (Goldenberg). SACN found "inconsistent evidence" at 12 to 24 months and "adequate evidence for no difference" at 24 months (SACN). The Cochrane review found little or no difference in HbA1c over 1 to 2 years (Naude 2022). The most likely reason is that people drift back towards their usual eating. Goldenberg's team found weight loss on very-low-carb diets depended on adherence (Goldenberg).

Can a low-carb or keto diet put type 2 diabetes into remission?

Possibly for some people, but the evidence is weak. At six months, 57% on low-carb diets reached HbA1c under 6.5% against 31% on control diets (risk difference 0.32, 95% CI 0.17 to 0.47; 8 trials, n=264). In plain terms, about 1 extra person in 3 got below the threshold. That figure included people still taking medicines. Using the stricter test, HbA1c under 6.5% and no medicines, the result was 16 v 13 per 100, which is not statistically significant. The authors estimate about 1 extra remission for every 20 people. At 12 months, remission off medicines was slightly lower on low-carb (RR 0.79, 0.36 to 1.73), based on just 2 trials (Goldenberg 2021).

The same review found remission was much less likely in trials that included people on insulin. The risk difference was 0.14 there, against 0.51 in trials without insulin users (Goldenberg). This matches DiRECT, which recruited only people diagnosed within the past 6 years who were not on insulin (Lean 2024). Remission is most realistic early in the disease.

Low-carb v low-calorie (DiRECT): which is better for remission?

No independent trial has compared them head to head, so we cannot say. What we can say is that the low-calorie route has the strongest randomised evidence. In DiRECT, 46% of the intervention group were in remission at 12 months, against 4% of controls (odds ratio 19.7, 95% CI 7.8 to 49.8). Mean weight fell by 10.0 kg against 1.0 kg, and quality of life improved (Lean 2018). Remission fell to 36% at 2 years. At 5 years, 13% of the extension group were in remission, and 26% of people in remission at year 2 were still in remission. Serious adverse events were less than half as frequent in the intervention group (4.8 v 10.2 per 100 patient-years) (Lean 2024).

When the NHS in England rolled out a similar programme, results were lower. Of 1,740 people who started the formula diet before January 2022, 55% completed the year. Mean weight loss among starters with complete data was 9.4 kg. Of those with two HbA1c results recorded, 27% met the remission definition. The authors note that remission rates and data completeness were lower than in the trial (Valabhji 2024).

What about Virta's "diabetes reversal" results?

Virta's programme combines a ketogenic diet with remote coaching and doctors who adjust medicines. Its 1-year study enrolled 262 people in the programme and 87 on usual care. The groups were not randomised. HbA1c fell from 7.6% to 6.3%, and weight fell by 13.8 kg. Insulin was reduced or stopped in 94% of users, and sulfonylureas were stopped entirely (Hallberg 2018, funded by Virta Health). At 2 years, use of glucose-lowering medicines other than metformin had fallen from 55.7% to 26.8% (Athinarayanan 2019, sponsored by Virta). At 5 years, 122 of the original 262 remained. Of these completers, 20% met the remission definition and 32.5% had HbA1c under 6.5% on no medicine or metformin only (McKenzie 2024, Virta authors).

These are seller-run, non-randomised results, so they cannot show the programme caused the change. The 5-year figures come only from people who stayed in, and people who stay are usually the ones doing well. The heavy medical support is also part of the package. Independent replication with randomisation is missing.

Does it help weight loss?

Only in the short term. Low-carb diets led to 3.46 kg more weight loss at six months (95% CI −5.25 to −1.67; n=882). By 12 months the difference was 0.29 kg in the other direction, which is effectively none (Goldenberg). SACN found "adequate evidence for no difference" in weight at 12 months or more (SACN).

Keto or a lower-carb Mediterranean diet?

In the Keto-Med crossover trial, 40 adults with prediabetes or type 2 diabetes spent 12 weeks each on a "well-formulated" ketogenic diet and on a lower-carb Mediterranean diet that kept legumes, fruit and whole grains. HbA1c improved on both, with no difference between them. The keto diet lowered triglycerides more but raised LDL. It also gave less fibre and lower intakes of three nutrients. Follow-up suggested the Mediterranean version was easier to keep up (Gardner 2022). The ADA cites this trial in its 2026 Standards (ADA 2026).

Risks and all side effects

LDL cholesterol

Results vary between people and between diets. Pooled trials showed no LDL difference at six months, with high certainty. At 12 months LDL was 0.14 mmol/L higher on low-carb diets (95% CI −0.00 to 0.28). That is above the authors' threshold for a clinically important change, but only borderline statistically (Goldenberg). In Keto-Med, LDL rose 10% on keto and fell 5% on the Mediterranean diet (Gardner). SACN found no LDL difference overall (SACN). Virta reported no significant LDL change at 5 years (McKenzie 2024, Virta authors). Averages can hide large rises in some people, so repeating a lipid test after a few months on the diet is sensible. SACN advises people on lower-carb diets to limit saturated fat (SACN).

Quality of life and sustainability

Quality of life was similar at six months. At 12 months it was worse on low-carb diets by a clinically important margin, although this was not statistically significant (Goldenberg). In Keto-Med, the keto diet appeared less sustainable after the trial (Gardner).

Other side effects

SACN found the most common adverse events across trials were gastroenteritis, nausea, vomiting and headaches. It found no difference from higher-carb diets over 3 to 6 months, and said longer-term effects are "currently unknown" (SACN). The BMJ review found no significant increase in total or serious adverse events, but rated this evidence low or very low certainty because few trials reported harms properly (Goldenberg). Very-low-carb diets that cut out fruit, legumes and whole grains lower fibre intake (Gardner). In DiRECT's formula-diet arm, one participant had biliary colic and abdominal pain judged possibly related to the intervention (Lean 2018).

All interactions with diabetes and other medicines

Who should avoid a very-low-carb diet

• People taking SGLT2 inhibitors, unless a specialist has agreed a plan (ADA 2026).
• People who are pregnant or breastfeeding, children, people with kidney disease, and people with or at risk of disordered eating. The ADA says very-low-carb plans are not currently recommended for these groups (ADA 2026).
• People with low insulin reserve, such as latent autoimmune diabetes in adults (LADA), low C-peptide or past pancreatitis. Regulators flag these as ketoacidosis risk factors with SGLT2 inhibitors (MHRA).
• People with type 2 diabetes who are not overweight. SACN found too little evidence to make any recommendation for this group (SACN).
• People from minority ethnic groups. These diets are not unsafe for them, but most trials were in predominantly White populations, so it is not known whether the effects apply equally (SACN).

Carbohydrate levels studied, and keeping it up

These are the levels used in research, not personal targets:

• BMJ meta-analysis: low-carb under 130 g a day (or under 26% of energy); very-low-carb under 10% of energy; trials lasted at least 12 weeks (Goldenberg).
• Cochrane: low-carb more than 50 g up to 150 g a day (or under 45% of energy); very-low-carb 50 g a day or less, compared with "balanced" diets of 45–65% carbohydrate (Naude).
• DiRECT: 825–853 kcal a day of formula for 12–20 weeks, then 2–8 weeks of stepped food reintroduction, then maintenance support (Lean 2024).

On food quality, SACN advises people who choose lower carb to include wholegrain or higher-fibre foods and a variety of fruit and vegetables, and to limit saturated fat. It recommends lower carb "as an effective short-term option (up to 6 months)" for adults with type 2 diabetes and overweight or obesity. It also says health professionals should support "any evidence-based dietary approach" that achieves long-term weight loss (SACN). The ADA suggests that eating fewer processed foods is an effective way to cut carbohydrate (ADA 2026).

On testing, the remission consensus recommends checking HbA1c at least 6 months after a lifestyle change and 3 months after stopping medicines, then at least yearly. People in remission should also keep having routine checks for diabetes complications (Riddle 2021).

Follow the money: who sells it and who funded the evidence

Who sells low-carb and remission programmes

Virta Health (San Francisco) sells a ketogenic diet with remote medical supervision to employers and health plans. In April 2021 it raised $133 million in a round led by Tiger Global, at a $2 billion valuation. It says it puts "100% of its fees at risk" on clinical and financial outcomes. Its press release used the language of "diabetes reversal", a term the 2021 consensus chose not to use (Virta press release; Riddle 2021).

Atkins is owned by The Simply Good Foods Company of Denver, Colorado, a Nasdaq-listed company that also owns Quest and OWYN. Net sales were $1.45 billion in the year to August 2025, with Atkins sales declining. The company describes Atkins as aimed at people following a low-carbohydrate lifestyle or "seeking to manage weight or blood sugar levels". Its annual report says that since 2016 it has advocated with the US federal government for "more inclusive guidance" in the Dietary Guidelines for Americans, including in meetings with the Department of Agriculture and the Department of Health and Human Services. It also says it has sponsored continuing-education courses for dietitians and other health professionals (Simply Good Foods 10-K, 2025).

Low-carb books. Australia's national science agency, CSIRO, publishes the CSIRO Low Carb Diet book. One of its authors co-wrote the BMJ meta-analysis and declared that he receives no personal royalties (Goldenberg 2021). On the low-calorie side, DiRECT co-lead Roy Taylor declared authorship of the book Life Without Diabetes in the remission consensus (Riddle 2021). He has also advised Fast800 (Valabhji 2024).

Formula-diet companies. Cambridge Weight Plan donated the DiRECT formula diet. Some DiRECT investigators were employees of, or shareholders in, Counterweight Ltd when the study began (DiRECT protocol). The 5-year paper discloses PhD fees and conference support from Cambridge Weight Plan for some authors, outside the submitted work (Lean 2024).

Who funded the trials

• DiRECT: Diabetes UK, as a Strategic Research Initiative (protocol). Diabetes UK's income in 2024 was £38.4 million, of which £27.93 million came from donations and legacies (Charity Commission). It also has corporate partners, including Novo Nordisk (Diabetes UK).
• Virta studies: Virta Health Corp. paid for the study and the publishing charges. Most authors were Virta employees with stock options (Hallberg 2018). Two co-founders were authors of the 2-year paper (Athinarayanan 2019). The 5-year paper is copyrighted to Virta Health and lists Virta affiliations, yet states that the authors have "no known competing financial interests" (McKenzie 2024).
• Goldenberg meta-analysis: partly funded by Texas A&M University, which had no role in the work. The senior author receives Texas A&M AgriLife funds for a separate project on saturated and polyunsaturated fats (Goldenberg 2021).
• Keto-Med: research gifts from private donors, plus NIH grants. The funders had no role (Gardner 2022).
• Cochrane review: some authors partly funded by UK aid; no conflicts declared (Naude 2022).
• Guideline bodies: The ADA pays for its Standards from general revenue, with no industry money (ADA). The organisation as a whole received $24.8 million from drug and device makers in 2023 (ADA disclosure). Many members of the remission consensus panel declared payments from drug makers or from makers of obesity-surgery devices such as Ethicon and Medtronic (Riddle 2021).

Documented integrity events in nutrition science

Sugar-industry documents show the Sugar Research Foundation paid for a 1960s literature review in the New England Journal of Medicine. The Foundation set the review's objective, contributed articles and received drafts. The review blamed fat and cholesterol for heart disease and played down sucrose, and the funding was not disclosed (Kearns, Schmidt & Glantz 2016). The analysts concluded that the industry "successfully cast doubt" about the hazards of sugar in the 1960s and 1970s. Low-carb advocates often cite this history. It shows that undisclosed funding can shape nutrition advice. It does not show that today's balanced-diet advice is wrong, and the same scrutiny applies to the companies that sell low-carb products and programmes.

We found no regulatory action against Virta or Atkins in the sources we reviewed. That is not the same as confirming there has been none.

Related research

• Type 2 Diabetes: Prevention and Management Evidence
• Diabetes and Heart Disease: Cardiovascular Risks, Treatment Purposes and Warning Signs
• Diabetes Supplements: Benefits, Limits and Safety
• Heart-Healthy Dietary Patterns: DASH, Food Choices and Medicine Safety
• High Cholesterol: Assessment, Treatment, Supplement Risks and Evidence Funding
• Statins and Other Cholesterol-Lowering Medicines: Options and Safety
• Berberine: Glucose Evidence, Formulations and Safety
• Psyllium Husk: Cholesterol, Blood Sugar, IBS and Safety
• Stevia: Forms, Intake Limits and Human Evidence
• Sucralose: Evidence on Glucose, Gut Microbiome and Safety
• Monk Fruit: Forms, Blood Sugar Evidence and Safety Limits

Frequently asked questions

Sources and funding notes

What we could not verify: We could not find an independent randomised trial comparing a low-carb diet head to head with a low-calorie formula diet for remission. We could not find any trial of low-carb diets measuring heart attacks, complications or death in type 2 diabetes. We could not confirm the head-office country of Cambridge Weight Plan or Counterweight from the sources we opened, so we have not stated it. We did not find Atkins brand-level sales figures in the filing we read, only company totals and a statement that Atkins is declining. We did not find Virta's total funding raised to date, current valuation or revenue. We did not find a published diet-specific warning for metformin, GLP-1 receptor agonists or DPP-4 inhibitors. The NHS programme evaluation reports remission only for participants with complete HbA1c data, and many lacked it. We did not review the imaging studies behind the "liver and pancreas fat" theory of remission.

1. Scientific Advisory Committee on Nutrition. Lower carbohydrate diets for adults with type 2 diabetes (2021); GOV.UK page — UK government advisory committee; joint working group with Diabetes UK, British Dietetic Association and Royal Colleges.
2. Goldenberg JZ et al. Efficacy and safety of low and very low carbohydrate diets for type 2 diabetes remission. BMJ 2021 (PubMed) — systematic review; partly funded by Texas A&M University; one author linked to the CSIRO Low Carb Diet book.
3. Naude CE et al. Low-carbohydrate versus balanced-carbohydrate diets for reducing weight and cardiovascular risk. Cochrane 2022 — Cochrane review; authors partly funded by UK aid (READ-It).
4. Riddle MC et al. Consensus Report: Definition and Interpretation of Remission in Type 2 Diabetes. Diabetes Care 2021 — ADA-convened panel with EASD, Diabetes UK and the Endocrine Society; extensive author industry disclosures.
5. Lean MEJ et al. Primary care-led weight management for remission of type 2 diabetes (DiRECT). Lancet 2018 — randomised trial; funded by Diabetes UK.
6. Lean ME et al. 5-year follow-up of DiRECT. Lancet Diabetes Endocrinol 2024 — extension study; Diabetes UK; author ties to Counterweight and Cambridge Weight Plan disclosed.
7. Leslie WS et al. DiRECT study protocol. BMC Fam Pract 2016 — funding: Diabetes UK; formula donated by Cambridge Weight Plan.
8. Valabhji J et al. Early findings from the NHS Type 2 Diabetes Path to Remission Programme. Lancet Diabetes Endocrinol 2024 — NHS England service evaluation; no funding declared.
9. Gardner CD et al. Keto-Med randomized crossover trial. Am J Clin Nutr 2022 — Stanford; private research gifts and NIH grants.
10. Hallberg SJ et al. Effectiveness and safety of a novel care model at 1 year. Diabetes Ther 2018 — non-randomised; funded by Virta Health; Virta-employed authors.
11. Athinarayanan SJ et al. 2-year non-randomized trial. Front Endocrinol 2019 — sponsored by Virta Health; founders among authors.
12. McKenzie AL et al. 5-year effects of a continuous remote care model. Diabetes Res Clin Pract 2024 — Virta-affiliated authors; copyright Virta Health.
13. American Diabetes Association. Section 5, Standards of Care in Diabetes—2026; Introduction and methodology — ADA general revenue.
14. ADA. Revenue received from pharmaceutical/device makers, 2023 (unaudited draft) — ADA disclosure.
15. MHRA Drug Safety Update, April 2016: SGLT2 inhibitors and diabetic ketoacidosis — UK medicines regulator.
16. Jardiance (empagliflozin) prescribing information, DailyMed — FDA-approved label, written by the manufacturer.
17. Tougaard NH et al. Very low carbohydrate diet and SGLT-2-inhibitor: double jeopardy in relation to ketoacidosis. BMJ Case Rep 2019 — Danish hospital case report; no competing interests.
18. Kearns CE, Schmidt LA, Glantz SA. Sugar industry and coronary heart disease research. JAMA Intern Med 2016 — UCSF; NIH grants and a family fund donation.
19. Virta Health press release: $133M Series E (April 2021) — company statement; self-interested.
20. The Simply Good Foods Company, Form 10-K for fiscal 2025 — owner of Atkins; securities filing.
21. Charity Commission: Diabetes UK accounts; Diabetes UK: Novo Nordisk partnership — charity regulator and charity's own statement.

Educational research reviewed 6 October 2026. This article is not medical advice; talk to a qualified clinician or pharmacist before starting, stopping or combining treatments.