Half Or More From Carbohydrate: Type 2 Diabetes, Remission, And What To Know Before Carnivore

In 1982 Britain’s diabetes charity told people who could not handle glucose to get half or more of their energy from carbohydrate. Thirty years later a furious patient in Southport, and then two professors in Newcastle and Glasgow, showed the disease could go away. What they found, how far the evidence carries you, and what has to happen to your medication before you try carnivore.

In 2012 a woman with type 2 diabetes walked into Norwood Surgery in Southport with normal blood sugar and a temper.

She had found the low carbohydrate diet by herself. She had lost weight, she wasn’t hungry, she felt wonderful, and her sugars were normal. Her GP had written to her about the metformin she had stopped taking, meaning to tell her off. She told him instead that he had never explained to her that bread, breakfast cereal, rice and chocolate biscuits all turn into sugar, and that if she had known, she would have been doing this for ten years.

Then she told him she was beginning to wonder “whether you’ve got O-level biology”.

Her GP was David Unwin. He had spent the first twenty-five years of his career treating type 2 diabetes the way he had been taught, as a condition that only goes one way, adding a drug, then another, then another. In all that time he had never seen it go into remission. He said later that she frightened him, that he didn’t blame her, and that he decided the thing to do was listen.

She was right about the biology. She had picked the wrong culprit, or at least the most junior one. The advice her GP had been handing out went back thirty years, to a committee of the British Diabetic Association, and to a decision that the people least able to handle carbohydrate should get half or more of their energy from it.

The older history, from a Scottish army surgeon’s meat diet for diabetes in 1797 onwards, is in the forgotten history of carnivore nutrition. This story starts in 1981.

Half or more

For most of the twentieth century a diabetic diet meant a carbohydrate ration. As late as 1978, a British team testing a new high carbohydrate plan against the old way of doing things described the old way, in the title of their paper, as “standard low-carbohydrate dietary advice”.

Within three years the English-speaking diabetes world had turned the ration upside down.

The American Diabetes Association went first, in 1979, raising carbohydrate for insulin-dependent patients to 50 to 60% of calories. The Canadian Diabetes Association followed with 45% or more for every diabetic patient and no upper limit at all. In July 1981 the nutrition subcommittee of the British Diabetic Association finished its draft, and in 1982 the charity published it. Half or more of the energy in a diabetic diet was to come from carbohydrate, preferably from fibre-rich, unprocessed foods, with sugar kept down.

The charity had been founded in 1934 by the novelist H. G. Wells and the physician R. D. Lawrence, both of whom had diabetes. It became the British Diabetic Association in 1954, and it has been Diabetes UK since June 2000.

The reasoning was cardiac. People with type 2 diabetes were dying of heart disease well above the normal rate, dietary fat was the chief suspect, and so fat was to come down for every diabetic patient, whether their blood fats were raised or not. Take fat off a plate and something has to fill the space. The committees filled it with starch.

The second influence was the fibre hypothesis of the 1970s, which held that diets built on unrefined starchy staples protected against heart disease, bowel cancer and diabetes. It carried an unspoken assumption that more carbohydrate must be good for you too, and the same thinking had already shaped America’s Dietary Goals of 1977, which led to its official dietary guidelines in 1980.

Not everyone was convinced. The Stanford endocrinologist Gerald Reaven published an editorial in Diabetologia in 1980 under the title “How high the carbohydrate?”, and he warned that lifting the old ration might be taken as a licence to eat the carbohydrate foods that compromise diabetic control.

Even the new advice’s friends had their doubts. Reviewing the shift in the same journal in 1982, three researchers who were broadly in favour of the fibre idea wrote: “Sheer weight of support does not indicate the rightness of a manoeuvre.”

Nine years later the subcommittee updated its advice for the 1990s. It kept the 1982 principles and put carbohydrate at about 50 to 55% of energy, mostly from complex sources. In the same summary it admitted that some high carbohydrate diets had been shown to “worsen glucose control” and blood fats. For overweight patients, it said, the most important aim remained eating less energy.

Nine names are listed on that update. One of them is Michael Lean.

Ten teaspoons

Back in Southport, David Unwin wanted to offer his diabetic patients what had worked for the woman who had told him off. His partners said no. The practice was already short of appointments, and they wanted him seeing sick people rather than running something new.

He told his wife, Jen, that the surgery wasn’t paid to do it. She told him she thought she had married a doctor, asked why anyone needed paying to do a good thing, and offered to run the sessions herself, unpaid, on Monday nights. She is a consultant psychologist. The practice nurse volunteered too. They began with 18 patients, swapping recipes and cooking together.

The teaching was simple. Starch is the plant’s way of storing sugar, so bread, rice, potatoes and breakfast cereal all end up as glucose in the blood. Unwin translated it into a unit his patients already understood, the teaspoon of sugar. By his reckoning a small bowl of boiled rice, 150 grams of it, digests to the equivalent of about ten teaspoons. Out went the cereal, rice, bread and potatoes. In came meat, fish, eggs, full-fat dairy, green vegetables, berries and nuts.

The surgery kept the numbers from the start. Between 2013 and 2021, 186 of the 473 people on the practice’s type 2 register chose the diet, and they stayed on it for an average of 33 months. Their median weight fell from 97 kg to 86 kg. Their median HbA1c fell from 63 mmol/mol to 46, from well inside the diabetic range to below it. Their systolic blood pressure fell by 12 mmHg on average, and their LDL cholesterol went down.

Ninety-four of them reached drug-free remission: an HbA1c below 48 mmol/mol with no diabetes medication at all. That was 51% of the people who chose the diet, and 20% of every person with type 2 diabetes on the surgery’s books.

Time mattered most, along with the starting HbA1c. Of the people who started within a year of diagnosis, 77% went into remission. Of the people who had lived with diabetes for more than 15 years, 20% did.

The biggest falls in blood sugar came where nobody expected them. The patients who never reached remission, mostly people who had been diabetic for years and arrived with the highest HbA1c, saw their median HbA1c drop from 75 to 54.

The drugs bill was lower too. Norwood spent £4.94 per patient per year on diabetes drugs, against £11.30 at the practices around it, and in the year to January 2022 it spent £68,353 less than the area average.

It was one practice, the patients chose themselves, and there was no control group, all of which the authors say in their own paper. They also wrote down something the low carbohydrate camp tends to skip. Nobody at Norwood went into remission without losing some weight, and in the authors’ view, weight loss by any means can probably bring remission about.

The man who had shown why was the paper’s fifth author.

Three months of shakes

In 2011 Roy Taylor, a professor of medicine at Newcastle University, put 11 people with type 2 diabetes on 600 calories a day for eight weeks, mostly diet drinks with some non-starchy vegetables, and scanned their insides as they went. Almost a third of their liver fat drained out within a week, and pancreas fat fell over the following weeks. Their livers began responding to insulin again, and the cells that make insulin went back to working normally. The paper put it down to “dietary energy restriction alone”.

In Taylor’s account, type 2 diabetes in most people begins with too much fat in the liver. The liver stops responding to insulin properly, and the excess fat it exports settles in the pancreas, where it stops the insulin-making cells doing their job. Take enough fat out and both organs can recover, provided the insulin cells still can.

In 64 people from the big trial that followed, liver fat fell from 16% to 3.1%. His later work concluded that losing the fat is necessary and not always enough: remission also depends on whether the insulin cells are still able to recover. That fits Norwood’s numbers on duration. The longer someone has had the disease, the less there is left to recover.

Diabetes UK paid for the big test, and Taylor ran it with a Glasgow professor of human nutrition. His name was Michael Lean, the same Michael Lean whose name is on the 1991 update, the one that said the most important aim for overweight diabetic patients was to eat less energy. DiRECT took that sentence as far as it would go.

Between 2014 and 2016 the trial recruited 306 people aged 20 to 65 through 49 GP practices in Scotland and Tyneside, all diagnosed within the previous six years and none on insulin.

Half the practices carried on as normal. In the other half, every diabetes drug and every blood pressure drug was stopped on the first day under the trial’s supervision, to be restarted if readings rose, and the patients lived on a formula diet of 825 to 853 calories a day for three to five months. Food was then brought back over two to eight weeks, with support to keep the weight off.

At 12 months, 68 of the 149 people in the diet group were in remission, 46%. In the comparison group it was six.

The formula came from the Cambridge Weight Plan, and its make-up is printed in the trial paper: 59% carbohydrate, 26% protein, 13% fat, 2% fibre.

At 825 to 853 calories a day, that comes to roughly 120 to 125 grams of carbohydrate, which sits inside the 50 to 130 grams a day that Diabetes UK now counts as a low carbohydrate diet. The trial that put remission on the NHS kept the 1982 ratio and served less than half the calories of an ordinary day. At that portion size, nobody can eat much of anything, and the old fight between counting calories and cutting carbohydrate goes quiet.

When food came back, it came back at about half carbohydrate, and then the trial watched what happened.

At two years, 36% of the diet group were still in remission. In a five-year extension, 11 of the 85 people who had kept receiving support were in remission, 13%, and of those who had been in remission at two years, about a quarter still were. The weight drifted back, and Lean and Taylor’s own conclusion is that sustained weight loss is what drives remission.

The NHS buys the shakes

NHS England didn’t wait for year five. In 2020 it piloted its own version, the Type 2 Diabetes Path to Remission Programme, run jointly with Diabetes UK: twelve weeks of soups, shakes and bars at 800 to 900 calories a day, then food brought back, all inside a 12-month programme with coaching.

Its criteria include a diagnosis in the last six years and a BMI over 27, or over 25 for people from Black, Asian and other ethnic groups. More than 30,000 people have started it. In 2024/25 more than 13,000 enrolled, nearly double the 6,401 of the year before, and it now runs in every local health area in England.

NHS England’s headline is that one in three who completed it put their diabetes into remission. The evaluation behind it, published in The Lancet Diabetes & Endocrinology in 2024, is more particular about who that third were. Of the 1,740 people who started the shakes before January 2022, 55% completed the programme. Among the starters who had two HbA1c results on record, 27% were in remission at a year. The 32% belongs to the 450 completers with the same records.

That is a real result, delivered at scale, by a health service that had long treated the disease as a one-way street.

What the NHS still says

On Diabetes UK’s figures for 2024/25, there are over 4.7 million people in the UK with a diabetes diagnosis, about 90% of them with type 2, and almost 1.3 million more who have type 2 and don’t yet know it. In 2023/24, 71 million prescription items were issued for treating diabetes in England, 21 million more than in 2015/16. The disease costs the NHS an estimated £10.7 billion a year, 60% of it spent on complications.

For everyone who isn’t on the shakes, the diet advice has barely moved.

NICE’s guideline for type 2 diabetes, amended again this year, still tells doctors to encourage adults with the condition to follow “the same healthy eating advice as the general population”: high-fibre, low-glycaemic-index carbohydrate such as fruit, vegetables, wholegrains and pulses, low-fat dairy, oily fish, and a check on saturated fat. That recommendation is dated 2009.

When the government’s Scientific Advisory Committee on Nutrition reviewed low carbohydrate diets with Diabetes UK in 2021, the yardstick it measured them against was the standing UK advice that about half of energy should come from carbohydrate. The ratio from 1982 is still the benchmark.

The NHS website, last reviewed in February 2025, tells people with type 2 diabetes to eat fruit, vegetables, wholemeal bread, oats, chickpeas and lentils, and not to eat too much processed food high in salt and sugar, listing “bacon, chips, cheese, sweets and biscuits”. Two of the five contain next to no carbohydrate.

The same page says that people who need medicine are usually offered metformin and an SGLT2 inhibitor, one of the flozins, first. Along with insulin, the sulphonylureas and the related meglitinides, the flozins are the drugs that turn a very low carbohydrate diet into a medical decision.

Before you try carnivore

Is remission real?

Yes. And in Britain the word has an agreed meaning: an HbA1c below 48 mmol/mol, holding for at least three months, with no glucose-lowering medication at all. That definition comes from an international consensus published in 2021. What HbA1c measures, and where the lines sit, is in the blood tests article.

Remission lasts as long as whatever produced it lasts. In DiRECT, as the weight crept back over five years, most of the remissions went with it.

The checks carry on as well. The NHS checks HbA1c in people with type 2 diabetes every three to six months, checks weight, cholesterol, blood pressure and kidneys at least once a year and feet every year or two, and offers eye screening every one or two years. People in remission are meant to keep receiving that annual care. Turn up to all of it.

Your odds depend most on how long you’ve had the disease. Norwood’s paper calls the first year after diagnosis a window of opportunity, and both DiRECT and the NHS programme only take people diagnosed within six years. If you were diagnosed twenty years ago, remission is less likely and the diet still matters. The largest falls in HbA1c at Norwood came in people who never reached remission, and five of its patients got there from a starting HbA1c above 90.

What does cutting carbohydrate do that cutting calories alone does not?

Every remission in these British studies came with fat loss. Nobody at Norwood reached remission without losing some weight, and DiRECT’s verdict is that sustained weight loss drives it. Anyone who tells you carbohydrate restriction works by some route that has nothing to do with body fat is ahead of the evidence.

Taking the carbohydrate out still does things a calorie count does not.

It brings blood glucose down straight away, because less glucose is arriving. A 2019 prescribing guide in the British Journal of General Practice notes that blood glucose typically falls substantially when someone starts a low carbohydrate diet, which is why its dose changes happen at the start.

In the short term, it clears fat from the liver faster. In a two-week trial of 18 people with fatty liver, a carbohydrate-restricted group and a calorie-restricted group lost about the same weight, and the carbohydrate group cleared about twice as much liver fat. The details are in the blood tests article. If Taylor is right that type 2 diabetes starts in the liver, that is the organ you want emptied first.

It runs on ordinary food. Norwood’s patients ate meat, fish, eggs and full-fat dairy, nobody handed them a sachet, and the paper records that many of them were surprised not to feel hungry.

And it keeps working when remission doesn’t come, as Norwood’s longest-standing diabetics showed.

What can the evidence not tell you yet?

It can’t tell you how long the carbohydrate advantage lasts. The 2021 government review found benefits for HbA1c, fasting glucose and triglycerides for up to six months, judged the evidence too weak to recommend a lower carbohydrate diet for longer than that, and found no consistent evidence that people on lower carbohydrate diets lost more weight. Diabetes UK’s position statement calls lower carbohydrate diets “a short-term effective option” for adults with type 2 diabetes who carry extra weight.

It can’t tell you how low most people can go and stay there. A 2021 review in the BMJ pooled 23 trials and 1,357 people.

At six months, in the eight trials that reported it, 57% of people on low carbohydrate diets were in remission, counted as an HbA1c below 6.5% whether or not they took medication, against 31% on comparison diets, but when remission was counted strictly without medication the difference shrank and stopped being statistically significant. Twelve-month data were sparse.

The very low carbohydrate diets, under 10% of calories, lost less weight at six months than the moderate ones, and the authors traced that to people not sticking to them.

It can’t tell you what happens over decades. Both reviews dealt in HbA1c, weight and blood fats over months, and neither could speak to heart attacks, strokes or deaths. On cholesterol they disagreed: the government review found no difference in total or LDL cholesterol at any point, while the BMJ review saw LDL rise at 12 months by an amount large enough to matter clinically, though it was not statistically significant. Your GP is likely to raise it, so read the cholesterol article before the lipid results come back.

How far does it carry you towards carnivore?

Carnivore sits at the far end of the same line. NICE defines a very low carbohydrate diet as 20 to 50 grams of carbohydrate a day, and a diet of meat, fish, eggs and butter comes in at the bottom of that range or below it. Everything the guidance says about very low carbohydrate and ketogenic diets applies to you from the first day.

The logic of the British story points the same way. If the trouble is a body that can’t cope with the glucose arriving, the least glucose leaves it the least to cope with. Where the insulin resistance comes from in the first place is the subject of the metabolic dysfunction article. And if you’ve been told red meat itself causes type 2 diabetes, that claim is answered in the dangers article.

The evidence for carnivore itself is thin. The largest piece is a social media survey of 2,029 adults who had eaten a carnivore diet for at least six months, run in 2020 by a Harvard-led team and published in 2021.

Those with diabetes, type 1 or type 2, reported a median fall in HbA1c of 0.4 percentage points, a fall in BMI of 4.3, and widespread cuts to their diabetes medication. All of it was self-reported by people who chose to answer, and anyone who tried the diet and gave up before six months was excluded by design. It tells you plenty of people live on the diet happily. It cannot tell you what your HbA1c will do.

There is no controlled trial of a carnivore diet in type 2 diabetes to point to. Carnivore takes the one lever every British study pulled, less glucose arriving, to its limit, and the risks of the first fortnight rise with it. A very low carbohydrate diet is where blood sugar falls fastest, where drug doses have to change most, and where NICE wants a flozin discussed before you begin.

The ordinary first-month symptoms are in the first 30 days on carnivore and the side effects article. For a diabetic on medication, the shakes and sweats of a low blood sugar can be mistaken for them, and only a meter tells the two apart.

What happens to your tablets and insulin?

If you take any of these, speak to your GP, diabetes nurse or pharmacist before your first day without carbohydrate:

• gliclazide or any other sulphonylurea, or repaglinide
• insulin, of any kind
• a flozin: dapagliflozin, empagliflozin, canagliflozin or ertugliflozin, alone or combined with another drug in one tablet

Doses usually need changing from day one. Do not stop or change any diabetes or blood pressure medicine on your own.

That 2019 guide was written for prescribers adjusting diabetes drugs for a low carbohydrate diet, and its authors were two GPs, one of them David Unwin, with a consultant endocrinologist, an American medical professor and a pharmacist. This is what it suggests prescribers do.

Sulphonylureas, such as gliclazide, and the related meglitinides, such as repaglinide, should be reduced or stopped when the diet starts. A first cut of at least half is usually appropriate, with further cuts as blood glucose readings come down.

For insulin, halving the daily dose at the start is appropriate in most cases, or a smaller cut of around 30% where HbA1c is very high, with further reductions guided by blood glucose readings. Some people come off insulin altogether, over days or months.

Some must never stop it completely: people who make little insulin of their own, including some who have a slow-onset autoimmune diabetes that was diagnosed as type 2. That is more likely if you weren’t overweight when you were diagnosed, and it is found in some people who have had diabetes for a long time. Cutting their insulin too far sends their blood sugar very high and can lead to ketoacidosis.

The flozins carry a risk of ketoacidosis in anyone short of their own insulin, on any diet, and the guide warns that “SGLT2i-induced ketoacidosis may occur with a normal blood glucose”. A normal reading on the meter while you are seriously ill is exactly what makes it dangerous.

The guide suggests that in the community it may be appropriate for most patients to stop the flozin when the diet starts. NICE lists a very low carbohydrate or ketogenic diet among the risks to sort out before anyone starts a flozin, and tells doctors to warn people already taking one not to start such a diet without discussing it first, because the diet would raise their risk of ketoacidosis and the drug may need to be suspended. That decision belongs to your prescriber.

The ordinary ketosis of a carnivore diet is a normal physiological state. Ketoacidosis is a medical emergency.

Metformin is safe to continue, according to the guide, and so are the GLP-1 drugs such as semaglutide, which some people are later able to stop. The gliptins and acarbose are usually stopped because they do little once carbohydrate is low, and the glitazones because of concerns about their long-term risks.

Blood pressure tablets may need adjusting too, because cutting carbohydrate tends to lower blood pressure. In one earlier audit Norwood took a fifth of its patients’ blood pressure drugs away, and their pressure still fell. If you feel faint or dizzy when you stand up, get your blood pressure checked.

Anyone on a drug that can cause a hypo should have enough testing strips to check their blood sugar while the doses come down. The guide says so plainly.

Two emergencies are worth knowing by sight. A hypo shows up as shaking, sweating, sudden hunger, a pounding heart or confusion: treat it at once with something sugary, such as a small glass of fruit juice or glucose tablets, keep some in the house, and tell your diabetes team, because it usually means a dose needs to come down. Ketoacidosis on a flozin, or soon after stopping one, shows up as feeling or being sick, stomach pain, fast and deep breathing, unusual sleepiness or a sweet smell on the breath: get urgent medical help, even if your blood sugar reading is normal.

What to say to your GP or diabetes nurse

Book the appointment before the diet starts, and take these with you.

“I want to cut my carbohydrate right down, to almost none, well below 50 grams a day, from this date. Can we plan my medication before I start?” A date gives both of you something to plan around.

“Can we use the 2019 BJGP guide on adapting diabetes medication for a low carbohydrate diet?” It gives your prescriber a published starting point for most diabetes drugs.

“I take a flozin. NICE says I shouldn’t start a very low carbohydrate diet on it without discussing it with you. What do you want me to do with it?” The answer is theirs to give, and you need it before day one.

“Can I have enough testing strips while my doses come down, and who do I ring if my readings go low?” Get a name and a number before you start.

“What should I do about my blood pressure tablets if I start feeling dizzy when I stand up?”

“Can we check my HbA1c at three months, and if I’m under 48 at least three months after stopping all diabetes medication, can it be recorded as remission?” Then keep every check after that, eye screening included.

“Am I eligible for the Path to Remission programme?” If you were diagnosed within six years, know what the NHS offers before you choose.

The happy post

Norwood’s group sessions moved onto Zoom during the pandemic, and the link went out to every person with type 2 diabetes in the practice. Patients were sent graphs of their weight and blood results, which the reception staff called the happy post.

Unwin says he owes a lot to the woman who started it.

In eight years of offering patients a straight choice between the diet, with the surgery’s help, and medication for life, not one of them, the paper says, chose the medication.