Guide

Carbohydrate counting answers how much insulin. It says nothing about when, and the timing question is often the one causing the post-meal spike.

Rapid-acting insulin doesn't start working for about 15 minutes and peaks somewhere around 60 to 90. Fast carbohydrate peaks around 45. Dose at the first bite and you've arranged for the glucose to arrive first and the insulin to turn up after the peak has already happened.

Everything below is about the concept. The actual numbers — how long, how much, whether at all — belong to you and your care team, because they depend on your insulin, your ratios, and your body.

The two curves

Picture them on the same axis. Glucose from a high-GI meal climbs steeply from about 15 minutes and peaks near 45. Insulin from a mealtime dose barely registers before 15 minutes and peaks around 90.

The result of dosing at the first bite is a spike, followed by insulin still working hard after the food is finished — which is a low a few hours later. That's the classic spike-then-crash pattern, and it's a timing mismatch rather than a dosing error.

Pre-bolusing shifts the insulin curve leftward so the two overlap.

What changes the right lead time

Several things, and this is where glycemic load stops being an abstraction and becomes a dosing input.

Starting glucose. Higher pre-meal readings generally call for a longer lead; lower ones for a shorter one or none at all. Dosing 20 minutes ahead of a meal when you're already at the bottom of range is how a pre-bolus becomes a low.

The food's glycemic load. A high-GL meal — white rice, pasta, bread, anything refined — has a fast, tall curve that needs the head start. A low-GL meal spreads its glucose over hours, and pre-bolusing hard for it produces a low first and a rise later.

Fat and protein. Both delay the whole thing. Pizza is the standard example: fat slows gastric emptying, so glucose arrives late and keeps arriving. A single upfront dose is badly matched to that shape.

Anything slowing your stomach. Gastroparesis and GLP-1 medications both delay food arrival substantially, and both can turn a normal pre-bolus into a low. If either applies to you, this is specifically a conversation to have rather than an adjustment to try.

Where glycemic load earns its keep

This is the most practical application of GL I know of for anyone on mealtime insulin.

Two meals with identical carbohydrate counts can have completely different shapes. Forty grams from white rice arrives fast and tall. Forty grams from lentils and vegetables arrives slowly and low. Carb counting treats them identically, and they behave nothing alike.

Knowing which shape you're dealing with tells you how much lead time the meal wants, and whether the dose should arrive all at once or be spread out. That's information carbohydrate counting alone doesn't carry.

Split and extended dosing

For meals that arrive slowly — pizza, curry with cream, a long restaurant dinner — a single upfront dose is the wrong shape regardless of timing.

Pump users have extended and combination bolus features built for exactly this: part of the dose now, the rest delivered over the following hours. On injections, the equivalent is splitting into two doses separated by an hour or so.

Both approaches need working out with your care team and both benefit enormously from CGM data, because the pattern you're trying to match is only visible in a trace.

How to work out your own lead time

The method is straightforward and requires patience rather than expertise.

Pick a meal you eat often. Keep everything constant — same food, same portion, similar starting glucose — and vary only the pre-bolus timing across several occasions. Watch what happens between 45 and 120 minutes.

If you're spiking high and then falling, you likely need more lead time. If you're dropping before the food lands, you need less. Repeat each timing more than once, because a single result contains too much noise to act on.

A CGM makes this a few days of work. With finger sticks it's slower — a reading at 60 and 120 minutes gets you most of the way.

The 20-minute problem in real life

The technique assumes you know when the food is arriving, and much of life doesn't work that way.

Restaurants are the obvious case: dosing on arrival and then waiting 40 minutes for a kitchen is how a pre-bolus becomes an emergency. The workable version is dosing when the food is visible rather than when it's ordered.

Cooking at home is easier, since you control the timing — dosing when the pan goes on rather than when the plate lands is usually about right.

And meals with other people run long. A dinner that stretches over two hours with courses arriving unpredictably is a poor candidate for a single upfront dose, and a better one for splitting.

When not to pre-bolus

When you're already low or heading down. When you're not certain the food is coming — a restaurant that hasn't served yet, a child who may refuse the meal. When the meal is low-GL enough that there's no spike to get ahead of. And when illness, gastroparesis or a GLP-1 makes the arrival time unpredictable.

None of that is exotic. It's just the recognition that the technique matches insulin to an expected curve, and that when the curve is uncertain, so is the technique.

Raise it at your next appointment, particularly if your post-meal numbers are consistently high while your pre-meal numbers look fine. That specific pattern is the signature of a timing problem, and it's one of the more fixable things in day-to-day diabetes management.