My 120-Day Carb Experiment

What would it take?

120 days, three very different approaches to carbohydrates, and one key question:

“What actually happens when you change the balance of carbs, fat, and protein in the real world, day after day, while living with type 1 diabetes?”

As a diabetes dietitian, the “low-carb versus high-carb” debate was already everywhere. Patients asked me daily: Should I cut carbs? Should I try keto? The answer, as always, is it depends. But I wanted to know what it really looked like in practice. Not the theory, not the perfect-world Instagram version — what it felt like to live it.

So, I did. And here’s how it went.

Why I Did It

At the time, my goals were simple:

• Lose about 10% body fat
• See how carb restriction affects glucose, insulin use, exercise, and lipids
• Test whether I could realistically sustain lower-carb approaches

I’d been in good shape already, but I wanted to go further — leaner, more consistent, and armed with first-hand evidence I could share with patients.

The Setup

I kept some rules constant to make it a fair test:

• Approximately 2,500 kcal per day (a deficit of approximately 1,000 kcal)
• 10,000 steps every day
• Training: 4x strength + 2x HIIT sessions per week
• Alcohol out, hydration in
• Consistent bedtime and wake-up

Then I split the 120 days into three phases of approximately 40 days each.

Nutritional breakdown and lifestyle parameters

Phase 1: High-Carb, Low-Fat (Days 0–40)

This was my baseline. The diet was flexible, but I had to stay disciplined to keep calories down.

• Macros: 45–55% carbs, 30–35% protein, 15–25% fat.
• Strategy: Three consistent meals per mealtime. It worked. Count the carbs, dose insulin, and glucose stayed predictable.
• Challenges: Any shifts in carb/fat balance threw things off. Pre-bolus timing, post-meal walks, and occasional corrections were necessary.
• Exercise: Worked fine with careful planning.
• Social life: Pretty manageable — nothing radical needed.

This phase reminded me how much predictability matters. I could manage anything with insulin and structure — but it required a lot of micromanagement, Dynamic Glucose Management style.

Phase 2: Low-Carb, High-Fat (Days 40–80)

Here I cut out starches and leaned on fruit, vegetables, and higher fat intake.

• Macros: approximately 50% fat, 30% protein, 20% carbs.
• Insulin adjustment: I had to count 50% of protein as carb — otherwise boluses weren’t accurate.
• Benefits: Way less need for pre-bolus precision or corrective walking. Meals were steadier, glucose more predictable.
• Challenges: Shopping and cooking had to be consistent — no room for last-minute changes.
• Exercise: Easier to manage glucose, but I noticed HIIT performance start to slip toward the end.
• Social life: Reasonable with some swaps (more steak and veg, fewer chips and bread).

This felt like the “sweet spot” of balance: lower stress, steadier glucose, but still functional for everyday life.

Phase 3: Keto (Days 80–120)

This was the strictest — and, honestly, the hardest.

• Macros: 60–70% fat, 30–35% protein, 5% carbs.
• Insulin: Counted only 25% of protein as carb. Boluses were minimal, and glucose was incredibly stable.
• Challenges: Constipation, electrolyte imbalance, hydration. Needed magnesium and potassium supplements.
• Exercise: Gym performance dropped hard. High-intensity was a struggle.
• Social life: Very tough. Eating out required planning ahead or just saying no.

KETO and blood ketone measurements

Although the last 40 days were called KETO, it is actually misleading. I measured my blood ketone levels every day — they never went above 0.2 mmol/l.

Only when I dropped my protein intake from 1.9 g/kg to 1.0–1.2 g/kg (100–120 g) did I get ketones in the nutritional ketosis range of 0.6–3.0 mmol/l.

The suggested benefits of having blood ketone levels of 0.6–3.0 mmol/l are:

• Reduced hunger.
• Anti-inflammatory effect in the brain by using ketones rather than glucose as fuel.
• Providing the brain with around 30% of its fuel, potentially increasing cognitive sharpness.

Ketones are only made in the liver and kidneys, so for the brain to be getting a consistent supply, blood levels need to consistently sit at 0.6–3.0 mmol/l.

Why ketone levels didn’t reach the target range

There is one main reason: the effect of protein on insulin and glucagon.

Amino acids from protein digestion cause the alpha-cells in the pancreas to release glucagon. To prevent too much glucagon from being released and liberating glucose from the liver, a healthy pancreas also releases insulin from the beta-cells in response to protein ingestion. The insulin directly inhibits the alpha-cells from releasing too much glucagon. This works effectively for people with a healthy pancreas if protein intake is kept within limits:

• For non-active people: 1.2 g/kg (0.5 g/lb)
• For very active people: 2.0 g/kg (0.9 g/lb)

If protein intake goes above this, glucagon is released in large amounts, and more insulin is released from the beta-cells. The high level of insulin reaches the liver and kidneys where it turns off ketone production. Therefore, if protein intake is too high, nutritional ketosis of 0.6–3.0 mmol/l is not possible, and the proposed benefits cannot be accessed.

This challenge is even greater for the person with type 1 diabetes. Because the person with type 1 diabetes infuses insulin into the peripheral tissues rather than secreting it directly from the beta-cells of the pancreas, the direct inhibition of glucagon release from the alpha-cells is less effective. A higher insulin level is needed to achieve the same result — and that higher circulating insulin is even more effective at stopping the liver and kidneys from producing ketones.

Therefore, the protein level cut-offs need to be lower for people with type 1 diabetes who want to achieve nutritional ketosis (0.6–3.0 mmol/l):

• For non-active people: 0.8–1.0 g/kg (0.3–0.4 g/lb)
• For very active people: 1.2–1.5 g/kg (0.5–0.7 g/lb)

So in simple terms, if a person with type 1 diabetes wants to get into nutritional ketosis, they need a lower protein intake and higher fat intake. This may not be ideal for maintaining lean muscle mass in an energy-restricted phase.

Bottom line: if you want a true ketogenic diet with blood ketones consistently at 0.6–3.0 mmol/l, you need to keep protein at less than 1.5 g/kg (0.7 g/lb) and have a fat intake of at least 75% of total energy. And you need to test your blood ketones — not guess.

How I Insulin Dosed for Each Diet

Insulin dosing for the different diet types created a unique challenge.

One thing is certain: carbohydrate counting alone did not work across the different macronutrient splits.

Taking a deep dive into the research and science is beyond the scope of this write-up. That information can be found in the Ultimate Guide to Insulin Dosing.

Results

I wanted to see what would happen over the 120 days to:

• Diabetes management and insulin requirements
• Metabolic markers
• Exercise performance
• Body composition

Diabetes management and insulin requirements

The experiment clearly shows: the lower the daily carbohydrate intake, the easier my blood glucose was to keep in target.

This was mainly due to never having a high amount of bolus insulin in the system at any one time. The lower the bolus insulin, the less chance of a quick drop in glucose. The less carbohydrate, the less chance of spikes.

Put simply: smaller inputs, smaller outputs.

Interestingly, I enjoyed the majority of the benefit in glucose management by dropping from 345 g a day to 131 g a day. There was only a slight further benefit by dropping from 131 g to 16 g a day.

Metabolic markers

Blood pressure dropped progressively and finished in the optimal range, which is what would be expected with improved body composition.

Highlights from the lipid results:

• Triglycerides remained stable and at a low-risk level throughout.
• HDL increased progressively with energy restriction and body fat loss.
• There was a big increase in HDL when carbohydrate decreased — in line with evidence showing insulin reduces HDL production by the liver.
• All risk ratios (TC:HDL, HDL:LDL, TG:HDL) improved as energy was restricted and body fat was lost.
• The TG:HDL ratio reduced significantly — this ratio is regarded as a strong predictor of coronary heart disease risk in large cohort studies.

The concerns:

• Total cholesterol was already at 5.9 mmol/l before the experiment. My paternal grandfather died of cardiovascular disease and my father had a recent total cholesterol of 9.0 mmol/l — there are genetic elements at play here. This result ultimately led to a diagnosis of Familial Hypercholesterolaemia and the start of lipid-lowering therapy. Total cholesterol did drop with weight loss and energy restriction in the first 40 days of HCLF, but then increased as carbohydrate decreased and fat increased — reaching a potentially concerning 8.6 mmol/l by day 120.
• LDL cholesterol increased consistently as carbohydrate intake decreased and fat increased — undoubtedly influenced by the undiagnosed Familial Hypercholesterolaemia. This is why testing lipids before embarking on any radical dietary regimen is important.

Exercise performance

Key observations:

• A progressive drop in strength across the dieting phase — to be expected with prolonged dietary restriction. During KETO, in the third and fourth sets of compound lifts such as deadlifts and squats, I felt completely empty. Not a feeling I enjoyed at all.
• The HIIT results were the most striking. By the end of 120 days, I could only complete 70% of my usual work. Some of this will have been due to prolonged dieting, and some likely due to reduced glycogen storage. I believe the lack of carbohydrate impacted performance well above and beyond energy restriction alone — but I acknowledge I could be biased, and re-running the experiment in reverse order would be the only way to confirm this.
• It was also my cricket season during the LCHF and KETO periods. During LCHF I felt ok in the field for three hours. During KETO I felt very lethargic and it seriously impacted performance. Again, in my view, primarily due to carbohydrate restriction rather than energy restriction alone.

Body composition

Highlights:

• Fat mass dropped consistently from approximately 15 kg to approximately 5 kg.
• Lean mass held fairly steady, thanks to consistent strength training.
• Waist circumference shrank consistently across all three phases.
• Energy restriction is the driver of body composition change — not macronutrient breakdown.

What I Took Away

Looking back, this experiment wasn’t just data — it was lived reality. And that’s the bit that matters most when you’re trying to translate science into actual life with diabetes.

Some of my biggest personal takeaways:

• Building a handful of consistent, repeatable meals is the key to predictable glucose — regardless of macro split.
• Protein matters. On lower-carb plans, it still impacts glucose and needs to be accounted for in insulin dosing.
• Glucose management tends to improve with lower carbs, but exercise performance at high intensities may suffer at the extremes.
• Keto isn’t keto unless protein is kept below the threshold and ketones are consistently measured. Most “keto diets” out there are really just low-carb.
• Lipids need monitoring. Energy restriction improved some ratios, but high fat intake can raise cholesterol significantly — particularly in those with underlying genetic risk.
• Fat loss came from the calorie deficit — not from carbs versus fat as a category.