The Keto Diet and Type 2 Diabetes: What It Does, and Who It Isn't Safe For

Keto arrives with better numbers than most diet advice, and that is exactly why it deserves a careful reading rather than an enthusiastic one. People do see their fasting readings drop within days. Some do come off medication. Those reports are not exaggeration — the trials broadly agree with them. What the reports usually leave out is which medication you were on when you started, and what the same trials found twelve months in.

Keto is not simply "low-carb"

The two get used interchangeably and they are not the same thing. Diabetes UK defines a low-carb diet (opens in a new tab) as cutting carbohydrate to under 130g a day — for most Pakistani households that means dropping to one or two rotis a meal and treating rice as an occasional dish rather than a daily base. It is a real change, but a recognisable one.

Keto goes considerably further. The carbohydrate floor is low enough that the body runs short of glucose to burn and switches to breaking down fat instead — the state Cleveland Clinic describes as ketosis (opens in a new tab). In practice that means no roti, no rice, no naan, no potato, no fruit beyond a few berries, and no daal in any meaningful quantity. The American Diabetes Association's Standards of Care (opens in a new tab) group anything under 26% of total energy from carbohydrate as "very low-carbohydrate"; keto sits at the bottom of that band.

That last row is the one people underestimate. A low-carb plate can be served from the same degchi as everyone else's dinner. A keto plate usually cannot.

What the evidence actually shows

It shows a genuine effect, and it shows that effect fading. A meta-analysis of fifteen studies in type 2 diabetes (opens in a new tab) found ketogenic diets lowered HbA1c by 1.45% against control diets, alongside 2.67 kg more weight loss. Diets that were merely low-carb managed 0.27%, and their weight difference did not reach statistical significance. On those figures keto is not a milder version of low-carb — it is doing something categorically stronger.

Two caveats sit inside that result, and the authors raise both. The ketogenic findings varied enormously between studies, and no trial compared keto against low-carb directly — each was measured against whatever control diet its own researchers chose. The authors also close by weighing the glycemic and weight benefits against increased cardiovascular risk and an unfavourable blood lipid profile, which is not a footnote when you already have diabetes.

The ADA's reading of the wider literature adds the timeline. Carbohydrate-restricted patterns do reduce A1C and the need for glucose-lowering medication — in the short term, under six months, with less difference between eating patterns once you pass a year. Diabetes UK reaches the same conclusion in one sentence:

“There's no evidence that following a low-carb diet is any more beneficial in managing diabetes than other approaches in the long term.”

None of that makes keto a bad choice. It makes it a choice whose advantage is front-loaded — real and useful in the first months, converging with everything else after that. Which means the question worth asking before you start is not whether it works. It is whether you can still be doing it in two years, and what your plan is if you can't.

One medication makes this genuinely unsafe

If you take an SGLT2 inhibitor — dapagliflozin, empagliflozin, canagliflozin, and the combination tablets built around them — do not start a keto diet without speaking to your doctor first. Diabetes UK is explicit that you may need to stop the medication, because these drugs raise the risk of diabetic ketoacidosis when carbohydrate intake is cut.

Insulin and sulfonylureas — glimepiride, gliclazide, glipizide, glibenclamide — carry a different problem. They lower blood sugar on a schedule that assumes you are eating the carbohydrate you normally eat. Remove it and the dose becomes too large for the meal, and hypoglycemia follows. This is manageable, but only by your doctor adjusting the dose, and ideally before the first low rather than after it. Diabetes UK notes the same thing happens again as you lose weight: the dose that fitted you in month one may not fit you in month four.

If you manage type 2 diabetes with diet alone or with metformin, hypoglycemia is much less of a concern and the conversation with your doctor is a shorter one. It is still worth having.

Who this isn't for

Diabetes UK does not recommend carbohydrate restriction for people with type 1 diabetes, on the grounds that there is no strong evidence it is safe or beneficial for them. It also advises against it for children, where the evidence shows an effect on growth. If you are pregnant, breastfeeding, living with kidney or liver disease, or have a history of disordered eating, this is a decision to make with a clinician who knows your history rather than from an article.

The part that decides it in a Pakistani home

Everything above is medical. What usually settles whether keto lasts is domestic. Roti, rice and daal are not merely carbohydrate here — they are how a household eats together, what gets cooked in one pot, what gets sent to a neighbour, what appears at every wedding and every Eid lunch. Keto asks you to step outside that at more or less every meal, and to keep doing it while sitting at the same table.

Some people manage it comfortably, particularly where the cooking is already meat-and-vegetable heavy and someone at home is willing to plan around it. Others find they are eating separately at every meal within a fortnight and quietly stop. Neither outcome says anything about discipline. The ADA's own position is that there is no single ideal macronutrient split for everyone with diabetes, and that a plan has to be built around culture, preference, access and sustainability — not just around numbers. A low-carb approach you keep for a decade will outperform a keto plan you keep for five weeks, and the trials at the one-year mark broadly agree.

If you decide to try it

1. 1Settle the medication question first — before the first meal, not after the first bad reading. Bring your full prescription list, including anything you take for blood pressure or heart failure, since SGLT2 inhibitors are prescribed for those too.
2. 2Test more often in the first two weeks than you normally would, and log it. Falling readings are the expected result; they are also the signal that a dose needs revisiting.
3. 3Ask for a lipid profile before you start and again at three months. The lipid effect is the part the trials are most uneasy about, and it is invisible without a test.
4. 4Pick a date — three months, six months — to sit down and honestly assess whether this is still working and still liveable, rather than drifting into abandoning it.
5. 5Decide in advance what you fall back to. Reverting to how you ate before the diet is the outcome that undoes the benefit; a sustainable lower-carb version of your normal food is the one that keeps it.

The first fortnight is the part worth logging

Cutting carbohydrate sharply changes your readings fast, and those early days are what tell you and your doctor whether a dose needs adjusting. Log your readings and meals in Diatic so the pattern is there to show, rather than remembered.

Start logging (opens Diatic on Google Play in a new tab)