I needed to settle the carbohydrate debate swirling mind.
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:
~2,500 kcal per day (a deficit of ~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 ~40 days each.
Nutritional breakdown and Lifestyle Parameters
Usual diet before experiment HCLF 40 days LCHF 40 days KETO 40 days Average daily carbohydrate intake (g) 350g (3.6g/kg) 345g (3.5g/kg) 131g (1.4g/kg) 16g (0.2g/kg) % total energy from carbohydrate 44% 50% 20% 3% Average daily fat intake (g) 140g (1.5g/kg) 55g (3.5g/kg) 140g (1.5g/kg) 186g (2.0g/kg) % total energy from fat 35% 20% 50% 67% Average daily protein intake (g) 185g (1.9g/kg) 185g (1.9g/kg) 185g (1.9g/kg) 185g (1.9g/kg) % total energy from protein 21% 30% 30% 30% Daily energy intake 3,500kcal 2,500kcal 2,500kcal 2,500kcal Supplements Creatine 3g/d Multi-vit& min 1xd Caffeine 400mg/g Whey Protein 90% Creatine 3g/d Multi-vit& min 1xd Caffeine 400mg/g Whey Protein 90% Creatine 3g/d Multi-vit& min 1xd Caffeine 400mg/g Whey Protein 90% Creatine 3g/d Multi-vit& min 1xd Caffeine 400mg/g Whey Protein 90% Slo-Mag (300mg Mg, 500mg Ca) Potassium 2g/d Sodium 2g/d Average hours sleep per night 7.5 hours 7.5 hours 7.5 hours 7.5 hours Average time exercising per week 300 minutes 300 minutes 300 minutes 300 minutes
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.
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 Managemen t 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: ~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: 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 0.2mmol/l.
Only when I dropped my protein intake from 1.9g/kg to 1.0-1.2g/kg (100-120g) did I get ketones in the nutritional ketosis range of 0.6-3.0mmol/l.
The suggested benefits of having blood ketones levels of 0.6 – 3.0mmol/l are:
Reduced hunger.
Anti-inflammatory effect in the brain by using ketones rather than glucose as fuel.
Providing the brain with 30% of its fuel, therefore, increasing cognitive sharpness.
Plus much more, depending on your source of information.
Ketones are only made in the liver and kidneys, so for the brain to be getting a consistent supply, the blood levels should be consistently 0.6-3.0mmol/l.
There is one main reason why my ketone levels were not getting 0.6-3.0mmol/l.
But first we need to look at how a healthy pancreas operates.
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, the “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 to a limit of:
For non-active people: 1.2g/kg (0.5g/lb)
For very active people 2.0g/kg (0.9g/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 makes it way to the liver and kidneys where it turns off ketone production. Therefore, if protein intake is too high, nutritional ketosis of 0.6-3.0mmol/l is not possible, and the proposed benefits cannot be enjoyed.
This issue is even more challenging for the person with Type 1 diabetes. This is because the person with type 1 diabetes infuses insulin into the peripheral tissues, it is not secreted directly from the beta-cells of the pancreas .
This means the direct inhibition of glucagon release from the alpha-cells is not as effective. Therefore, an even higher insulin level is needed to achieve the same result of stopping liver glycogen liberation. This higher level of circulating insulin is even more effective at stopping the liver and kidneys, producing ketones.
Therefore, the protein level cut-offs need to be lower for people with type 1 diabetes wanting to achieve nutritional ketosis, blood ketones of 0.6-3.0mmol/l, are:
For non-active people: 0.8-1.0g/kg (0.3-4g/lb)
For very active people 1.2-1.5g/kg (0.5-.7g/lb)
The diagram summarizes this:
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 is, if you want a true ketogenic diet that gets blood ketones of 0.6-3.0mmol/l consistently. You would need to keep protein at less than 1.5g/kg (0.7g/lb) and have a fat intake that is at least 75% of total energy intake.
You also need to test your blood ketones, and not guess!
How did I insulin dose for the different macro splits of the diets?
Insulin dosing for the different diet types created a unique challenge.
One thing is for sure! Carbohydrate counting alone did not work for 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 my:
Diabetes control and insulin requirements.
Metabolic markers.
Exercise Performance.
Body Composition.
Diabetes control and insulin requirements
Before change After HCLF 0-40 days After LCHF 40-80 days After KETO 80-120 days Predicted HbA1c% from sensor glucose readings 6.5% 6.3% 5.1% 5.5% Average blood glucose (mmol/l) 8.5mmol/ 8.0mmol/l 6.2mmol/l 6.1mmol/l Standard Deviation 3.5mmol/l 3.1mmol/l 1.8mmol/l 1.3mmol/l Number of hypos (<3.5mmol/l) each week 5 4 2 1 Total daily insulin (TDD) (units) 45units 38 units 25 units 18 units Total daily basal insulin (units& %TDD) 18units (40%) 15 units (39%) 15.5units (62%) 11 units (61%) Total daily bolus insulin (units& % TDD) 27units (60%) 23 units (61%) 9.5 units (38%) 7 units (39%)
The experiment clearly shows.
The lower the daily carbohydrate intake, the easier my blood glucose was to stay in target!
This was mainly due to never having a high amount of bolus insulin in my system at any one time. The lower the bolus insulin in your system, the less chance of a quick drop in glucose. Also, the less carbohydrate, the less chance of spikes in blood glucose.
Put simply, smaller inputs, smaller outputs!
Interestingly, I enjoyed the majority of the benefit in blood glucose control by dropping from 345g a day to 131g a day. There was only a slight further benefit in diabetes control by dropping from 131g to 16g a day.
Metabolic markers
Before After HCLF 0-40 days After LCHF 40-80 days After KETO 80-120 days Blood pressure (systolic/diastolic) 140/80 135/75 130/70 123/70 Triglycerides (TG) Ideal <1.7mmol/l 1.3mmol/l 1.0mmol/l 1.0mmol/l 1.0mmol/l Total cholesterol (TC) Ideal <5.0mmol/l 5.9mmol/l 4.8mmol/l 7.3mmol/l 8.6mmol/l Low Density Lipoprotein (LDL) Ideal<3.0mmol/l 4.1mmol/l 3.2mmol/l 4.6mmol/l 5.4mmol/l High Density Lipoprotein (HDL) Ideal>1.0mmol/l 1.3mmol/l 1.5mmol/l 2.2mmol/l 2.7mmol/l TC:HDL (Risk: low <3.5, medium 0.5, high >0.7) 4.5 Medium 3.2 Low 3.3 Low 3.2 Low HDL:LDL (Risk: low >0.4, medium 0.3, high <0.2) 0.3 Medium 0.4 Low 0.47 Low 0.5 Low TG:HDL (Risk: low <0.87, medium 1.74, high >2.62) 0.77 Low 0.66 Low 0.45 Low 0.37 Low
My blood pressure dropped progressively and finished in the optimal range, which is what would be expected with improved body composition.
The highlights from the lipid results:
The positive changes in lipid profile:
TG remained stable and at low risk level.
HDL increased progressively with energy restriction and body fat loss.
There was a big increase in HDL when carbohydrate decreased, which is in line with the evidence showing insulin reduces HDL production by the liver.
All the risk ratios (TC:HDL, HDL:LDL,TG:HDL) improved as energy was restricted and body fat lost.
TG to HDL ratio reduced significantly with energy restriction and fat loss. TG to HDL ratio is the strongest predictor of coronary heart disease in large cohort studies.
The concerns:
TC at 5.9mmol/l before the experiment. My paternal grandad died of cardiovascular disease and my dad had a recent TC of 9.0mmol/l. So there are some genetic elements at play here (this results led to a diagnosis of Familial Hypercholesterolemia, and taking lipid lowering therapy) ! TC did drop with weight loss and energy restriction in the first 40 days of HCLF. Then increased as carbohydrate decreased and fat increased to a potentially concerning 8.6mmol/l by day 120.
LDL cholesterol increased consistently as carbohydrate intake decreased and fat increased. No doubt due to undiagnosed Familial Hypercholesterolemia. This is why you should always test before embarking on any radical dietary regimen!
Exercise Performance
Before change After HCLF 0-40 days After LCHF 40-80 days After KETO 80-120 days Back squat x 8 Reps 140kg 140kg 135kg 130kg Dumbbell Chest Press x 8 Reps 100kg 100kg 95kg 90kg Pull ups – strict form 12 11 10 9 HITT Session Using before experiment performance as 100% 100% 95% 85% 70%
Key points:
A progressive drop in strength across the dieting phase. This is to be expected with prolonged dietary restriction. Nothing concerning me here. But I did feel the drop in carbohydrate when following KETO in the third and fourth sets of compound lifts, such as deadlifts and squats. I felt completely gassed with nothing in the tank. Not a feeling I enjoyed, at all!
The High Intensity Interval Training (HIIT) is where the most interesting results were seen. By the end of the 120-days, I could only complete 70% of my usual work. Some of this will have been due to prolonged dieting, and some of this may have been due to the lack of glycogen storage in my muscles to create the quick energy needed. The only way to know for sure would be to re-run the experiment and do KETO first, the LCHF, the HCLF. If my performance was greater than 70% at the end, then it would be clear for me that the carbohydrate was a major factor. If I had to state what I believe, it would be the lower carb almost certainly impacted way above just energy restriction alone. But I could be biased.
It was also my cricket season during the LCHF and KETO periods. During the LCHF 40 days I felt ok in the field for three hours, but when I dropped to KETO, I felt very lethargic, and it seriously impacted my performance. Some of this will have been due to prolonged energy restriction, but in my opinion, again, it was mainly due to carbohydrate restriction.
Body composition
Before change After HCLF 0-40 days After LCHF 40-80 days After KETO 80-120 days Weight (kg) Average morning weight from at least 3 measurements 100kg 97kg 95.5kg 91.8kg Body fat from 7-site Skin Callipers in mm &percentage drop from previous 40 days (%) 98 72 (27%) 50 (30%) 37 (26%) Body Fat % – estimated from 7 site calliper measurements 15% 11% 8% 5.7% Fat mass (kg) & percentage drop from previous 40 days (%) 15kg 11kg (27%) 8kg (28%) 5.4kg (33%) Fat Free Mass (kg) 85kg 86kg 87.5kg 86.4kg Waist circumference(cm)& percentage drop from previous 40 days (%) 98cm 92cm (7%) 86cm (7%) 84cm (3%)
Highlights:
Fat mass dropped consistently from ~15 kg to ~5 kg.
Lean mass held fairly steady thanks to strength training.
Waist circumference shrank consistently.
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 must be counted.
Exercise control improves with lower carbs, but performance suffers at the extremes.
Keto isn’t keto unless protein is capped 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 can raise cholesterol significantly.
Fat loss came from the calorie deficit—not from carbs versus fat.
Final Word
The 120-day carb experiment showed me that you can make any macro split work—but the trade-offs are different. High-carb required constant micromanagement . Low-carb reduced stress but cut exercise performance. Keto simplified boluses but wasn’t sustainable socially, physically, or even physiologically without major sacrifices.
At the end of the day, the best diet is the one you can actually live with, so why not experiment to find your sweet spot?
References
https://jissn.biomedcentral.com/articles/10.1186/1550-2783-11-20
http://ajcn.nutrition.org/content/early/2016/07/05/ajcn.116.133561
https://jissn.biomedcentral.com/articles/10.1186/s12970-017-0174-y
https://www.ncbi.nlm.nih.gov/pubmed/16360350
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2664115/
http://www.ajconline.org/article/S0002-9149(04)00517-X/abstract