For patients · Reading time 10 minutes · Version 1.1 · Translation of the German page, medically reviewed on 8 October 2026 · the German version is authoritative · Diese Seite auf Deutsch
Eating with type 2 diabetes: what reverses it, and what only claims to
"Diabetes is for life" has not been true for a few years now. In many people it can be pushed back until sugar and tablets are history. Here is what that takes, in numbers from GP practices. Plus the oat days, a German method from 1902 that is being re-examined today, and a few findings hardly anyone knows.
The essentials in three sentences
- Type 2 diabetes arises from fat in the liver and pancreas. Ten to fifteen kilos less reverse it in more than half, most readily in the first years.
- How you lose weight is secondary. The Mediterranean diet is the foundation, formula and low-carb are tools, and the oat days help when a lot of insulin achieves little.
- Small things count: vegetables first, starch last, ten minutes of walking after eating, as little ready-made food as possible.
Before you change your diet
If you inject insulin or take tablets that lower sugar by releasing insulin (sulfonylureas such as glimepiride or gliclazide), each of these measures can cause hypoglycaemia: oat days, low-carb, formula diet, fasting. Then the dose is adjusted beforehand, not afterwards. Call before you start. With metformin, SGLT2 inhibitors or GLP-1 medicines alone the risk is low; you should still let us know.
1
Why the diabetes is reversible
What became clear about its origin in the last fifteen years, and what the DiRECT study made of it.
Too much fat in the wrong place
In most people, type 2 diabetes develops when the liver and pancreas are overloaded with fat. The liver then releases too much sugar, the pancreas too little insulin. Both are reversible as long as the insulin-producing cells are still there.
The personal fat threshold
It does not depend on the weight on the scales. Some develop diabetes at normal weight, others never despite being overweight. What matters is how much fat your own liver tolerates. Anyone who gets below that, usually with 10 to 15 kilos less, can push the diabetes back.
The earlier, the better
In the first six years after diagnosis, reversal succeeds most often. After that less often, but not never.
The DiRECT study: GP practices, not a hospital
298 people with type 2 diabetes for less than six years, without insulin, in British GP practices. One half received a formula diet of about 850 kilocalories a day for three to five months, then normal food again with support. The other half the usual treatment. "Without diabetes" meant: long-term sugar under 6.5 % and no diabetes tablets.
After 1 year
46 out of 100
without diabetes, without tablets
After 2 years
36 out of 100
without diabetes, without tablets
After 5 years
10 out of 100
without diabetes, without tablets
With usual treatment it was 3 out of 100 after two years. And of those who still weighed at least ten kilos less after two years, 64 out of 100 were without diabetes. After five years the group was on average still six kilos lighter; those who kept the weight off mostly stayed in remission. So the number falls not because the method wears off, but because the weight comes back.
How much weight, how much prospect
After one year in DiRECT, depending on how much the participants had lost. Slide.
57 out of 100 were without diabetes and without tablets after one year.
Bands: 0–5 kg 7 % · 5–10 kg 34 % · 10–15 kg 57 % · from 15 kg 86 %. Your entry stays on your device. Averages from one study; for you a direction.
2
The oat days
An old method for a particular case: a lot of insulin, still poor values. With instructions, with studies and with the warning that goes with it.
Where it comes from
Carl von Noorden described the oat cure in Frankfurt in 1902, long before insulin. It fell into oblivion and has been re-examined by German clinics in the last twenty years: in people with type 2 diabetes who have poor values despite high insulin doses.
How it works
Two to three days of almost nothing but oats: few calories, no fat, plenty of beta-glucan, a soluble fibre. Afterwards the body responds more sensitively to insulin again for weeks. Exactly why is not settled; that it works is shown by the studies.
The studies
Mannheim 2008
Few, small studies14 people with severe insulin resistance, two oat days in hospital: the insulin dose fell by 42 %, from 145 to 83 units a day, and was still down four weeks later. No control group.
Lammert 2008
Jena 2024
Few, small studies17 people on insulin, as outpatients, three oat days once or twice a month for three months: the insulin dose fell from 138 to 126 units, HbA1c by 0.3 points. Those who did two rounds a month gained more. Three in four kept it up, no severe hypoglycaemia.
Fiedler 2024, randomised pilot study
For whom, and what an oat day looks like
For people with type 2 diabetes and a high insulin requirement whose values are still not good. Not as a permanent diet but as a cure: two to three days, repeated every two to four weeks if needed.
| Morning | 60 to 80 g of porridge oats, cooked in water or thin vegetable stock, with a little cinnamon or a few berries |
| Midday | 60 to 80 g of porridge oats, savoury: in stock with herbs, leek or a small portion of vegetables |
| Evening | 60 to 80 g of porridge oats, with a little vegetable or unsweetened stewed fruit |
- In total about 180 to 250 g of porridge oats a day, around 800 kilocalories. Wholegrain rolled oats, not instant oats.
- No fat, no sugar, no milk. Up to 60 g of vegetables a day for flavour are allowed.
- Drink plenty: water, unsweetened tea, thin stock.
- Two to three days in a row. Afterwards eat normally, but sensibly.
- Agree the blood sugar plan with us beforehand: insulin is reduced clearly on oat days, otherwise hypoglycaemia threatens. Measure more often during the days.
3
Diets compared
Six ways, ordered by evidence. What matters is not the best way, but the one you actually take.
Mediterranean diet
Well establishedVegetables, pulses, nuts, wholegrain, fish, plant oil. In diabetes it lowers heart attacks and strokes, delays the start of tablets and is the form people keep up the longest. The foundation on which everything else builds.
Estruch 2018; Esposito 2009
Weight loss with a formula diet
Well establishedThree to five months of about 850 kilocalories a day from powdered meals, then a gradual return to normal food, with support from the practice. In DiRECT, after one year 46 % were without diabetes and without tablets, after two years 36 %, after five years 10 %. Those who kept the weight off mostly stayed in remission. In Germany this is not covered by statutory health insurance; the powders cost about 100 to 150 euros a month.
Lean 2018, 2019; Davies 2026
Low-carb
Some studies, small effectUnder 130 g of carbohydrate a day. After six months 57 % were in remission compared with 31 % on other diets, plus more weight loss and better blood fats. After twelve months the lead was gone, because most do not stick with it. Those who keep it up benefit; with insulin only with support.
Goldenberg 2021, 23 studies
Intermittent fasting
Some studies, small effect16:8 or 5:2. For losing weight as good as other calorie restriction, not better. For some easier to keep up because nothing has to be weighed. With insulin only with an adjusted dose.
several reviews
Plant-based
Some studies, small effectVegetarian or vegan lowers HbA1c and weight somewhat, mainly through less saturated fat and more fibre. The effect comes from what is added, not from what is left out.
several reviews
Old-school low-fat
Few, small studiesThe recommendation of the 1990s. Not harmful, but inferior to the other forms in studies. The "diabetic diet" of that era no longer applies.
guidelines since 2014
4
Newer findings hardly anyone knows
Small changes with measurable effect, from studies of the last ten years.
The order of eating
vegetables and protein first
The same meal, eaten the other way round: first vegetables and meat, fish or pulses, last bread, rice, potatoes. The rise in sugar after the meal is clearly smaller. In one study in prediabetes, 94 % kept it up over four months and ate more vegetables and protein of their own accord.
Shukla 2015, 2023
Ten minutes of walking after eating
−12 %
Ten minutes three times a day after the meal lowers the sugar after eating by 12 % more than the same 30 minutes in one go at another time. After dinner, where most of the carbohydrates usually are, by 22 %.
Reynolds 2016
Ultra-processed food
+40 %
People who eat a lot of ready meals, soft drinks, sausage and confectionery have a 40 % higher risk of developing diabetes. With existing diabetes it is the same lever: less of it.
Lane 2024
Sweeteners
no route to weight loss
Since 2023 the World Health Organization has advised against using sweeteners to lose weight: in the long run they do not help. For blood sugar they are better than sugar in the short term. The better solution is to drink less sweet, with or without sweetener.
WHO 2023
Fructose
no diabetic sugar
"Suitable for diabetics" used to mean: sweetened with fructose. Fructose burdens the liver and promotes exactly the liver fat that drives diabetes. These products were abolished in Germany in 2012; anything still called that is marketing.
German dietary regulation 2012
Coffee
allowed
Coffee, including decaffeinated, is linked with less diabetes in observational studies. Without sugar, and the milk counts as food.
several cohorts
5
The plate in everyday life
Without counting, without scales. This is what a meal looks like that does not push the sugar up.
Half: Vegetables and salad
Raw or cooked, every colour. Potatoes do not belong here, they are starch.
A quarter: Protein
Pulses, fish, poultry, egg, quark, tofu. Rarely red meat, hardly any sausage.
A quarter: Wholegrain or starch
Wholegrain bread, wholegrain pasta, brown rice, potatoes, oats. Less if you want to lose weight.
Plus: A tablespoon of good oil, water
Rapeseed or olive oil. To drink: water, tea, coffee without sugar. Alcohol on insulin only with food: it lowers sugar with a delay.
6
From other traditions, and from advertising
Further reading. Each item says how well it is supported by Western studies.
Cinnamon
Asia
Small studies show a small reduction in fasting sugar, larger ones do not. Harmless as a spice; cassia cinnamon in quantity burdens the liver. No substitute for anything.
Bitter melon
India, China
Contains insulin-like substances. The studies are small and contradictory. Fine as a vegetable, not needed as a capsule.
Apple cider vinegar
Europe
A tablespoon before eating slightly lowers the sugar rise in small studies. Drink it diluted, otherwise teeth and gullet suffer.
Oat days (Haferkur)
Germany, 1902
The only home remedy on this page with its own chapter above. Old idea, new studies, real effect, but only with support.
Chromium, magnesium, vitamin D as capsules
Supplements
In people with normal levels, no effect on sugar. A real deficiency is measured and treated at the practice, not guessed.
Gymnema and Ayurvedic mixtures
India
No robust studies, and some mixtures contained hidden diabetes medicines or heavy metals when tested. Please bring them in and show us.
7
Quick check: what will you take away?
Five questions for anyone who likes. No marks.
1. Can type 2 diabetes disappear again?
2. What are the oat days?
3. You eat rice with chicken and broccoli. What lowers the sugar after the meal?
4. When should you walk, if it is to help the sugar?
5. "Suitable for diabetics" on a packet means today:
No marks, no time pressure. Your progress is kept on this device only.
What this page cannot do
It applies to type 2 diabetes; type 1 is a different disease. It does not replace structured education, dietary counselling or adjustment of your medicines. Anyone taking insulin or sulfonylureas discusses every change with us beforehand. The percentages come from studies with many people; they show a direction, not a guarantee. This page is a translation of the German original, which is the medically reviewed version; where the two differ, the German text applies.
Version 1.1 · Translation of the German page, medically reviewed on 8 October 2026 · the German version is authoritative
Reviewed by Stefan Höhne, Facharzt für Allgemeinmedizin. Numbers, doses and emergency numbers are identical to the German version and checked automatically. If anything is unclear, the German text applies.
Sources: Lean and colleagues, Lancet 2018 and Lancet Diabetes Endocrinol 2019, Davies and colleagues 2026 (DiRECT, 1, 2 and 5 years); Taylor 2008 onwards (twin-cycle hypothesis, personal fat threshold); Lammert and colleagues 2008 (oat days, Mannheim); Fiedler and colleagues, J Clin Med 2024 (oat days, Jena); Goldenberg and colleagues, BMJ 2021 (low-carb); Estruch and colleagues, NEJM 2018 (Mediterranean diet); Shukla and colleagues 2015 and 2023 (order of eating); Reynolds and colleagues, Diabetologia 2016 (walking after meals); Lane and colleagues, BMJ 2024 (ultra-processed food); WHO 2023 (sweeteners). Medically responsible: Stefan Höhne, specialist in general practice (Facharzt für Allgemeinmedizin). This page is independent: no sponsor, no advertising, no brand names. All patient information (German)