Mealtime Insulin
Mealtime Insulin: Survive & Thrive
The plate is in front of you. Pasta, a bit of sauce, some bread on the side, and the part of the meal nobody talks about: the sum you do in your head before you eat. You have done it ten thousand times, and it still asks something of you every single time. This page is about doing that sum well enough to get on with your dinner, not perfectly. Mealtime insulin is the part of type 1 that most rewards a clear method and most punishes chasing the last gram.
Want to ask how a high-fat, high-protein meal changes the way your insulin lands?
Why mealtime insulin is the hard part
Basal insulin runs quietly in the background and forgives a lot. Mealtime insulin does not. It has to arrive at roughly the right size, at roughly the right moment, for food that varies from one day to the next. Get the timing wrong by twenty minutes and the same dose behaves differently. That is not a personal failing; it is the nature of the job. Carbohydrate counting is the skill that gets most everyday mixed meals into the right ballpark, and for the great majority of plates it is enough on its own.
The 130 percent ICR insight
Here is the thing your insulin-to-carbohydrate ratio quietly knows that you might not. It was never a pure carbohydrate ratio. A usual mixed meal carries some fat and some protein alongside the carbohydrate, and your ratio was calibrated against meals like that. In practice the ratio covers roughly 130 percent of the insulin a carbohydrate-only plate would need, because it is already paying for the fat and protein that normally come along for the ride. This is a teaching frame rather than a precise calculation, but it explains a lot: a genuinely lean, carbohydrate-only meal can run you low on your usual ratio, and a meal far heavier in fat and protein than usual can leave you high hours later. The ratio works well while a meal stays roughly within the usual envelope: about 40 to 55 percent carbohydrate, 10 to 20 percent protein, 20 to 40 percent fat. Step well outside that, and counting carbohydrate alone falls short even when the count is spot on.
Counting well enough, not perfectly
Accuracy has a point of diminishing returns, and it arrives earlier than most people think. For most everyday meals, counting to within about 10 grams of the true carbohydrate is good enough at a population level; chasing the last 5 grams rarely changes how the afternoon goes. When a food carries a label, the carbohydrate is the portion weight in grams multiplied by the carbohydrate per 100 grams, divided by 100. When there is no label, a handy-portion estimate gets you a usable starting weight: a fist of cooked pasta, a deck-of-cards portion of meat. Most non-starchy vegetables, leafy greens, broccoli, courgette, peppers, tomatoes, are small enough to ignore for counting; the starchy ones, potato, sweetcorn, peas, parsnip, butternut squash, carry real carbohydrate and earn a place in the sum. The meal-builder in the GNL app does this arithmetic for you so you can spend the attention on the meal, not the maths.
Pre-bolus timing, the biggest single lever
If you change one thing about mealtime insulin, change when you take it rather than how much. Taking rapid-acting insulin about 15 to 20 minutes before a typical mixed meal lines the insulin peak up with the glucose rise far better than dosing at the first bite. The dose can be right and still arrive late, and late insulin spends the first hour chasing a rise it should have met head-on. Faster-onset insulins and very high-GI meals can shorten that lead; high-fat or high-protein meals can lengthen it. Very young children with unpredictable intake are a separate conversation, and one for the diabetes care team.
The fat-and-protein meal, the late rise
Carbohydrate is the rise you can see coming. Fat and protein are the rise that turns up after you have stopped watching. Both raise glucose in the second half of the post-meal window, and the effect is additive: a meal high in both produces the largest, latest rise of all. In adult closed-loop work a high-fat dinner needed about 42 percent more insulin than a low-fat dinner with identical carbohydrate and protein, and still ran higher afterwards. Across four fat amounts, 20 to 40 grams of added fat needed only about 6 percent more insulin, while 60 grams needed about 21 percent. Protein under about 25 grams in a meal usually needs nothing extra on top of the carbohydrate dose; from roughly 50 grams upwards it drives a meaningful delayed rise. And the combination is the one that matters most: a high-fat, high-protein meal more than doubled the 5-hour glucose rise at the same dose, and reaching target needed about 65 percent more insulin, with a wide individual range of roughly 17 to 124 percent. That range is the honest part. The direction is reliable for everyone; the size is personal, and it is learned over time with your care team using your own CGM patterns.
Two ways to handle the takeaway-pizza meal
When a meal sits well outside the usual envelope, two compatible patterns help, and the choice is yours to make with your care team. The fat-protein unit method counts 100 kcal from fat and protein as one unit, dosed at the same ratio as 10 grams of carbohydrate and spread over an extended bolus. The GNL working rule, drawn from a Birmingham audit, suggests about 25 percent more insulin split evenly over roughly 2.5 hours for meals carrying 2.0 to 2.5 grams of carbohydrate per kilogram of body weight; outside that band the rule does not apply and the care team should be involved. Neither is better than the other. They are two routes to the same place.
Does an AID system handle all this for me?
Partly, and honestly. An automated insulin delivery system compensates when a meal is under-bolused, but it does not fully replace an accurate meal announcement. The residual gap usually shows up as more time above range after meals, not as more lows. The pre-meal bolus still does most of the work; the algorithm cleans up around it. Counting well and bolusing on time make the algorithm’s job easier, not redundant.
Survive & Thrive · Mealtime Series
Mealtime insulin at a glance
| Level | Name | What it describes (mealtime ICR practice) |
|---|---|---|
| 1 | Optimal | ICR calibrated with the care team; pre-bolus 15 to 20 minutes consistently; fat and protein patterns recognised and a split-bolus or FPU/KISS rule applied to outlier meals; CGM used to refine. |
| 2 | High | ICR calibrated; pre-bolus usual; some awareness of the fat and protein effect but no consistent outlier-meal rule yet. |
| 3 | Balanced | ICR set and broadly working for usual mixed meals; pre-bolus inconsistent; outlier meals handled ad hoc. |
| 4 | Gentle | Counting carbohydrate but ICR not yet confidently calibrated; dosing at the meal rather than ahead; fat and protein not yet considered. |
| 5 | Protective | Building the counting skill; using handy-portion estimates; safety-first, hypo-avoidant approach while learning; care team closely involved. |
These five levels describe how well-calibrated and how actively-managed your mealtime approach is, not how strong your ratio is. A stronger ratio is not universally better; pushing it harder trades post-meal numbers against hypo risk. Optimal is the picture when the ICR is dialled in, the pre-bolus is consistent, and outlier meals have a plan. Balanced is a ratio that works for usual meals while the timing and the outlier meals are still settling. Protective is the right and safe place to be while you are building the counting skill, with the care team close by. Wherever you are, the next step is one rung, not the top of the ladder.
In the GNL app
The carb-counting meal-builder
The live tool lives in the GNL app, where you can build a meal and keep it with you at the table. This is what it does and how it is put together; the working version is a tap away in the app.
How it works
- Add what is on your plate. Search a food, or type your own, and give the portion in grams.
- It totals the macros for you. Carbohydrate, fat and protein, summed across every item, drawn from the official UK food-composition dataset (McCance and Widdowson’s CoFID 2021, around 2,900 foods).
- It points out the patterns. When a meal is high in fat (from about 40 grams), high in protein (from about 40 grams), or high in both, it shows the population pattern and the timing that usually helps, in plain words.
- It checks the load against body weight. A large carbohydrate load for the body gets its own gentle note, and a lower body weight brings the pattern notes in sooner, because children are not small adults.
- It works out grams, never doses. The builder never tells you how many units to take; the dose is yours to agree with your care team.
This preview is an illustration of the app feature. The tool works out grams and points out patterns; it does not tell you how much insulin to take. The direction (fat and protein need more insulin, given later) is reliable; the amount for any one person is not, and is learned over time with your diabetes care team using your CGM patterns. Educational only. Not a prescription, not a medical device.
References
Counting accuracy and method
- Smart CE, et al. Diabet Med. 2009: carbohydrate quantity is the primary determinant of the mealtime insulin dose.
- Smart CE, et al. Diabetes Care 2009: counting to within about 10 grams does not meaningfully change post-prandial glucose.
- McCance and Widdowson’s The Composition of Foods Integrated Dataset 2021 (PHE / FSA): per-100g values used by the meal-builder.
Fat and protein effect
- Wolpert HA, et al. Diabetes Care 2013: high-fat dinner needed about 42 percent more insulin than low-fat at identical carbohydrate and protein.
- Bell KJ, et al. Diabetes Care 2015 (systematic review): fat and protein raise late post-prandial glucose; effect is additive.
- Bell KJ, et al. Diabetes Care 2020: dose-finding across fat amounts (about 6 percent at 20 to 40 g, about 21 percent at 60 g).
- Paterson MA, et al. Diabet Med 2017: protein below about 25 g needs no extra insulin.
- Smart CE, et al. 2013 (paediatric): fat and protein raise glucose in children; effect is relative to body size.
- Pankowska E, et al.: fat-protein unit method (100 kcal from fat and protein = one unit).
Pre-bolus timing
- Cobry E, et al. Diabetes Technol Ther 2010: a 20-minute pre-bolus improves post-prandial glucose versus dosing at the meal.
AID and the residual meal gap
- Closed-loop literature: the meal announcement still does most of the work; the residual gap shows as time above range, not lows.
Practical working rules
- GNL / Birmingham audit working rule (KISS): about 25 percent more insulin split over roughly 2.5 hours for 2.0 to 2.5 g carbohydrate per kg body weight.
The Glucose Never Lies® · Educational only. Not a prescription, not a medical device. · Full disclaimer