Type 2 remission: what the trials showed, and what the word does not mean

I have Type 1, and I am writing this page because the word "remission" keeps arriving at my door anyway. A relative reads a headline and asks why I have not tried the diet. A stranger on a plane wants to know about the soup. Someone with Type 2 who has just been diagnosed reads the same headline and hears "you can fix this," then reads a different one and hears "it is your fault if you do not." Both of those people deserve the actual evidence, in plain words, without anyone selling anything.

So here it is: what the trials did, what they found, who it worked for, what the word means, and the part about Type 1 that the headlines never include.

First, the thing this is not

Type 1 and Type 2 diabetes share a name, a blood test and almost nothing else that matters here.

In Type 1, the immune system destroys the cells that make insulin. Once they are gone they are gone. There is no reserve to coax back, no fat to shift off a pancreas that would restart it, and no version of the diet below that ends with someone with Type 1 not needing insulin. Anyone who tells a person with Type 1 otherwise is wrong, and if they are selling something they are dangerous.

In Type 2, the body still makes insulin, often plenty of it, but the body has stopped responding to it well, and over years the insulin-making cells get tired and overloaded. The research below is about that second situation, and specifically about whether taking the load off those cells, early enough, lets them recover.

This page uses the word "remission" only in its Type 2 sense from here on. If you have Type 1 and someone sends you this, you can stop reading at this paragraph and send it back to them.

What "remission" was defined to mean

Until 2021 the word meant different things in different papers, which is part of why the headlines are a mess. That year an international group convened by the American Diabetes Association, with the European and UK bodies, agreed one definition: an A1C below 6.5 percent (48 mmol/mol), measured at least three months after stopping all glucose-lowering medication, and holding.

Three parts of that matter more than they look.

"Without medication." If a drug is keeping the A1C down, that is good management, not remission. This matters enormously right now, because the newer weight-loss medicines produce A1Cs in the remission range for a lot of people while they take them. More on that below.

"At least three months." One good reading is a good reading. Remission is a state that holds.

"Remission," not "reversal" or "cure." The panel chose the word on purpose. Diabetes UK's explanation is the plainest: there is no evidence that remission is permanent, so nothing that implies permanence gets used. People go in and come out. The eye checks, foot checks and kidney checks continue, because the years of high glucose before remission still happened.

DiRECT: the trial everyone is quoting

The Diabetes Remission Clinical Trial ran in ordinary GP practices in Scotland and Tyneside, which is part of why it mattered: it was not a specialist clinic with hand-picked patients. People had been diagnosed with Type 2 within the previous six years, were not on insulin, and had a BMI between 27 and 45.

The programme was blunt. Stop the diabetes and blood pressure tablets on day one, under supervision. Replace all food with formula shakes and soups at around 825 to 850 calories a day for three to five months. Then reintroduce food over a few weeks with structured support to keep the weight off.

At one year, 46 percent of the programme group were in remission, against 4 percent of the usual-care group. Average weight loss was 10 kg.

At two years, 36 percent were still in remission. The relationship with weight was stark: of the people who had kept off 10 kg or more, 81 percent were in remission. Of those who had regained to under 5 kg lost, it was 5 percent. Remission tracked the weight, not the willpower.

At five years, published in 2024, with a lower-intensity support programme continuing for those who wanted it, 13 percent of the original programme group were in remission. Average weight loss had shrunk to around 6 kg. The trial's own authors put it simply: remission depends on keeping the weight off, keeping the weight off is hard, and ongoing support helps.

Three things I think are worth taking from that, as someone reading it from the outside:

The first year number is real and large. Nearly half, in primary care, with no drugs. That is not a fad result.

The five year number is also real, and it is the one the headlines leave out. Thirteen percent of people held remission for five years. That is thirteen people in a hundred who, five years on, are living without diabetes medication and with an A1C in the normal range. It is a genuine outcome for them and it is not the majority outcome.

And the decline is about the weight coming back, which is the most human thing in the study. The trial was not a test of character. It was a test of whether a metabolic state can be undone by a large early weight loss, and the answer was yes, for as long as the loss holds.

The NHS took it to scale

The NHS Type 2 Diabetes Path to Remission Programme is the DiRECT approach run as a service: twelve months, three of them on total diet replacement, then support. More than 25,000 people have been offered it.

The first real-world evaluation, published in 2024, found that of people who completed the programme and had the two A1C measurements needed to check, 32 percent were in remission, with an average loss of almost 16 kg. Counted across everyone who started and had the blood tests, it was 27 percent. Reaching the end of the programme was itself the hard part: a large share did not.

That is the honest shape of it. Roughly a quarter to a third of people who see it through reach remission. The rest do not, and many of the rest still improved their A1C and weight meaningfully without crossing the line.

It is not only about being very overweight

A smaller Diabetes UK-funded study called ReTUNE asked whether the same thing happens in people with Type 2 at a normal or slightly raised weight, with a BMI under 27. Twenty people, diagnosed within the previous six years, did a shorter version of the low-calorie diet. Fourteen of them, 70 percent, reached remission, after losing around 8 percent of their body weight.

The interesting part was where the weight came from. Scans showed fat inside the liver and pancreas dropping to the levels seen in people without diabetes. The theory behind all of this, from the Newcastle group that ran both trials, is that Type 2 in many people is a personal fat threshold being crossed in those two organs, and the number on the bathroom scale is only a rough proxy for it. Someone can be "normal weight" and still over their own threshold.

So "you are not overweight, so this cannot be your situation" is not what the evidence says. Neither is "lose weight and you will be fine." Both are too simple.

Who it tends to work for

Across these trials and the bariatric surgery literature, which has its own remission data going back further, the same factors keep appearing:

• Time since diagnosis. Shorter is better, consistently. In the surgery studies, people diagnosed under four years earlier held remission at ten years around 80 percent of the time; people with long-standing Type 2 far less. The insulin-making cells seem to recover when they have been overloaded for years, not decades.
• Not yet on insulin, and on fewer medications. Partly because those are markers of how far along things are.
• The size of the weight loss, and whether it holds. This is the one factor the person has any hand in, and the trials are clear that holding it is the hard part and support matters.
• Starting A1C. Lower to begin with, more likely to cross the line.

None of those is a verdict. They are the population-level odds, and they are why a diagnosis in the last couple of years is the moment clinicians get most interested in this conversation.

The weight-loss drugs complicate the word

Semaglutide and tirzepatide, the GLP-1 family, produce large weight losses and A1Cs in the remission range for many people with Type 2, and a 2025 observational study found a meaningful share of people on them met the A1C criterion. But by the agreed definition, remission requires being off glucose-lowering medication for three months, and these are glucose-lowering medications.

What happens when people stop them? The evidence so far says the weight tends to return and Type 2 tends to progress again, even after years of good control. There is not yet evidence that remission extends beyond the drug without it. So the honest description is: these medicines can produce excellent, sustained control while taken, and whether that becomes remission in the agreed sense is an open question that the trials are still running.

If you are on one of them and your numbers look like remission, that is a good outcome and it is worth exactly what it is. Stopping to "test" whether it holds is a decision for you and your clinician, with a plan, not an experiment to run alone.

Surgery, briefly

Bariatric surgery has the longest remission track record. In the STAMPEDE trial, five years after gastric bypass about 29 percent were in complete remission, and 23 percent after sleeve gastrectomy. Ten-year follow-ups of surgery cohorts find complete remission around 30 percent, partial remission on top of that, and relapse in a quarter to a third of people who had initially reached it, at roughly 10 percent a year. Same pattern as the diet trials: early diagnosis and sustained weight loss predict who holds it.

What coming out of remission means

It is common. It is expected. It is in the definition. Diabetes UK's guidance is that people in remission keep their annual checks precisely because coming out of it can be quiet, and because the complications risk from the years before does not reset.

If you reached remission and then did not hold it, the trial that everyone quotes had that happen to most of its participants between year one and year five. You were in the majority. The weight you lost and the years your A1C was lower were not erased by the number drifting back up.

How to read the next headline

Three questions settle most of them.

Which definition? Off medication for three months with an A1C under 6.5 percent, or something looser? A lot of "reversed their diabetes" stories are A1C on medication, which is good management wearing a bigger word.

How long was the follow-up? One year tells you whether the intervention works. Five years tells you whether remission holds. They are different questions with different answers.

Who was in the study, and who was left out? Recently diagnosed, not on insulin, under a weight ceiling, completed the programme. If you are outside those, the number does not describe you, and it was never claiming to.

And one more for my own people: does the article say which type? If it does not, it is about Type 2. If it says it applies to Type 1, close the tab.