Carbohydrate counting and dosing
Last reviewed November 17, 2026. OwieClear editorial team. Waiting for clinician review. Not medical advice. Talk with your doctor about your own care.
Dosing insulin for food is the part of type 1 diabetes that never becomes automatic. It is arithmetic performed several times a day, on estimates, for the rest of someone's life.
The arithmetic is learnable in an afternoon. What takes years is the judgement about everything the arithmetic does not capture.
Related pages: Type 1 diabetes
The bottom line
- Insulin-to-carbohydrate ratio sets the mealtime dose.
- A correction factor sets how much one unit lowers glucose.
- Insulin on board must be accounted for to avoid stacking doses.
- Fat and protein delay and extend the glucose rise.
- Structured education in carbohydrate counting improves control.
- How best to dose for high-fat and high-protein meals.
- Whether fully automated meal detection will replace counting.
- The best approach to dosing in young children with unpredictable intake.
The story, step by step
- 1. The ratio ()
- 2. Corrections ()
- 3. Insulin on board ()
- 4. Beyond carbohydrate ()
- 5. Eating with type 1 ()
What the evidence labels mean
Insulin to carbohydrate
What happens
One number, used constantly.
What the research shows
- The insulin-to-carbohydrate ratio states how many grams of carbohydrate one unit of rapid-acting insulin covers. A ratio of 1:10 means one unit per 10 grams.
- Ratios differ between people, and usually differ by meal in the same person, often stronger at breakfast because of morning hormones.
- Carbohydrate is counted in grams from labels, packet weights, apps and learned portion sizes.
- Fibre is sometimes subtracted from total carbohydrate, and sugar alcohols partially, though practice varies.
- Accuracy matters more at larger doses: a 20 percent error on 2 units is trivial, on 12 units it is not.
- Structured education programmes in carbohydrate counting improve control and quality of life, and are under-offered.
Why it matters
People consistently underestimate carbohydrate in restaurant meals and overestimate it in vegetables. Weighing food for a week recalibrates both.
What's still uncertain
- How much estimation error contributes to variability.
- Whether apps improve counting accuracy in practice.
Read more
- MedlinePlus: Diabetic dietGovernment
- NIDDK: Managing diabetesGovernment
- American Diabetes Association: standards of care resourcesNonprofit
Next: Corrections
Bringing a high down
What happens
The correction factor.
What the research shows
- The correction factor, or insulin sensitivity factor, is how much one unit of insulin lowers glucose, for example 1 unit dropping it by 50 mg/dL.
- A correction dose is calculated from the difference between current glucose and the target, divided by that factor.
- Corrections are usually combined with the meal dose rather than given separately.
- Sensitivity varies through the day and with illness, activity, stress and hormones, so one factor rarely fits every situation.
- Over-correcting is the commonest cause of a subsequent low, which then gets over-treated, producing a high, and the cycle repeats.
- Pump and smart-pen bolus calculators do this arithmetic and track the result, which reduces errors substantially.
Why it matters
The rollercoaster most people describe usually starts with a single over-correction rather than with anything they ate.
What's still uncertain
- Whether multiple correction factors by time of day help.
- How much bolus calculators reduce error in practice.
Read more
- NIDDK: Managing diabetesGovernment
- American Diabetes Association: standards of care resourcesNonprofit
- MedlinePlus: Blood glucoseGovernment
Next: Insulin on board
Why stacking goes wrong
What happens
Insulin given an hour ago is still working.
What the research shows
- Rapid-acting insulin lasts three to five hours. A dose given an hour ago still has most of its effect to come.
- Giving another correction before the first has finished is called stacking, and it is a leading cause of severe lows.
- Insulin on board is the amount still active, and bolus calculators subtract it automatically from any new correction.
- The practical rule without a calculator is to wait at least three hours between corrections unless glucose is rising fast.
- The same applies after treating a low: wait and recheck rather than eating more because the number has not moved yet.
- Closed-loop systems handle this automatically, which is one of the clearest safety gains they provide.
Why it matters
Impatience with a number that has not moved yet is behind a large share of the lows people experience. The insulin usually needs more time, not more insulin.
What's still uncertain
- The best minimum interval between corrections.
- How to teach insulin-on-board intuitively.
Read more
- NIDDK: Managing diabetesGovernment
- American Diabetes Association: standards of care resourcesNonprofit
- NIDDK: Low blood glucose (hypoglycemia)Government
Next: Beyond carbohydrate
Fat, protein and timing
What happens
Where the arithmetic stops working.
What the research shows
- Fat slows stomach emptying, so a high-fat meal produces a delayed and prolonged glucose rise, often three to six hours later.
- Protein in quantity is partly converted to glucose and also raises glucose late.
- This is why pizza, curries and large restaurant meals cause a normal reading at two hours and a high one at five.
- Strategies include splitting the dose, using an extended bolus on a pump, or adding a small later correction.
- Pre-bolusing, giving rapid insulin 10 to 20 minutes before eating, substantially reduces post-meal spikes and is one of the highest-value habits available.
- The order food is eaten in matters too: vegetables and protein before carbohydrate flattens the rise.
Why it matters
Pizza is the standard example for a reason. It is the clearest demonstration that carbohydrate counting alone cannot explain what glucose does.
What's still uncertain
- The best dosing strategies for high-fat meals.
- How much food order affects glucose in practice.
Read more
- MedlinePlus: Diabetic dietGovernment
- American Diabetes Association: standards of care resourcesNonprofit
- NIDDK: Managing diabetesGovernment
Next: Eating with type 1
No forbidden foods
What happens
Dose for it, do not avoid it.
What the research shows
- There is no type 1 diabetes diet. Any food can be eaten if it is dosed for.
- A generally healthy pattern matters for cardiovascular risk, which is the main long-term threat, but it is not a glucose strategy on its own.
- Low-carbohydrate eating reduces glucose variability for some people and requires careful insulin adjustment and attention to growth in children.
- Children and adolescents need adequate intake for growth, and restriction can do real harm.
- Disordered eating is substantially more common in type 1 diabetes, including deliberate insulin omission to lose weight, which is dangerous and under-recognised.
- A dietitian experienced in type 1 is part of standard care and widely under-used.
Why it matters
Framing food as allowed or forbidden damages control in both directions and is a particular risk in adolescence.
What's still uncertain
- How best to screen for disordered eating in type 1.
- The safety of low-carbohydrate approaches in children.
Read more
- MedlinePlus: Diabetic dietGovernment
- American Diabetes Association: standards of care resourcesNonprofit
- NIDDK: Managing diabetesGovernment