The Carbohydrate-to-Insulin Ratio, and the Rule That Estimates It

What the ratio is, and what it is not

The ratio answers one question: how many grams of carbohydrate does one unit of rapid-acting insulin cover in this person? Written 1:10, it means one unit per 10 g, so a plate holding 60 g of carbohydrate asks for six units. Written 1:15, the same plate asks for four. It is a personal constant in the same way that shoe size is personal — not derived from body weight or from any single measurement, but observed from how the person actually responds.

That last point is the one the arithmetic can obscure. The ratio is not a property of food and it is not a property of insulin. It is a property of a particular person's response to a particular insulin at a particular time of day, and it is established by testing: eat a known amount of carbohydrate, take the dose the current ratio implies, and see where the glucose lands afterwards. Every published estimating rule, including the one below, exists only to give that testing process a sensible place to begin.

Where the number 500 comes from

The 500 rule states: grams of carbohydrate covered by one unit = 500 ÷ total daily dose of insulin, counting basal and bolus together. The reasoning behind it is inverse proportionality. Someone who needs a lot of insulin every day is, by definition, less responsive to each unit, so each unit should cover fewer grams. Someone who needs little is more responsive, and each unit covers more. Total daily dose is the most readily available proxy for that responsiveness, so it becomes the denominator, and 500 is the numerator that made the resulting ratios line up with what diabetes teams were arriving at by trial.

It is worth being explicit about what kind of number 500 is, because it is easy to mistake for something more solid. It is not a physiological constant like a molar mass or a half-life. It is a calibration figure, fitted to clinical practice, and the fact that a second widely used version exists at 450 is itself the proof: if the number were a law of nature there would only be one of it. Both appear in diabetes education materials, and neither is presented in the primary literature as a substitute for individual titration.

How sharply the ratio moves

Because the rule is a division, the ratio changes fastest where the daily dose is smallest. Going from 20 to 30 units drops the ratio from 25.0 g per unit to 16.7 — a fall of 8.3 g for ten extra units. Going from 60 to 80 units, another twenty units, drops it only from 8.3 to 6.3, a fall of 2.1 g. The same absolute change in daily dose has four times the effect at the bottom of the range as at the top. Anyone estimating a ratio for someone on a small total daily dose is therefore working on a steep part of the curve, where a modest error in the daily dose figure moves the answer a long way.

Translate that into a plate of food and it becomes concrete. Take a meal holding 60 g of carbohydrate. At a total daily dose of 40 units the 500 rule says 12.5 g per unit, so 4.8 units. At 60 units it says 8.3 g per unit, so 7.2 units. The plate has not changed; only the estimate of the person's sensitivity has, by a plausible amount, and the resulting dose differs by 2.4 units. That is the entire argument for why an estimating rule is a starting point rather than an answer: the estimate itself has a spread wide enough to matter at every meal.

The 500 rule and the 450 rule

Both numbers are in circulation, and the historical distinction is the insulin they were built around. The 450 figure is the older one, associated with regular human insulin — the short-acting insulin that dominated before rapid-acting analogues, with a slower onset and a longer, flatter tail. The 500 figure came in with the analogues, whose sharper peak covers a meal more efficiently and therefore lets each unit account for slightly more carbohydrate. Some teams also keep 450 as the deliberately conservative option in anyone at higher risk of hypoglycaemia.

The size of the difference is easy to state. Since 450 ÷ 500 = 0.9, the 450 rule always gives a ratio that is 10% smaller, and because the dose is the carbohydrate divided by the ratio, it always asks for 11.1% more insulin for the same food. At a daily dose of 50 units that is 9.0 g per unit instead of 10.0, and a 60 g meal becomes 6.7 units instead of 6.0. Whether that difference matters depends entirely on the person, and deciding it is a clinical judgement made with someone who can see their readings — not a preference to be set on a website.

Three things the rule quietly assumes

First, it assumes the total daily dose is stable and known. It is the only input, so everything rests on it. But the daily dose moves with illness, with activity, with a change in basal, with a growth spurt in a child, with pregnancy, with steroids, with weight change. A ratio estimated from last month's daily dose is an estimate of last month's sensitivity, and it goes stale silently — nothing announces that the denominator has drifted.

Second, it assumes one ratio serves the whole day, and in many people it does not. Insulin sensitivity varies with the circadian rhythm of counter-regulatory hormones, which is why a good number of people who count carbohydrate end up using a different ratio at breakfast from the one they use at dinner — often a tighter one in the morning, when cortisol and growth hormone are working against the insulin. A single estimating rule cannot produce that structure, because it has only one input and returns only one number. Discovering it takes testing, meal by meal, with a team.

Third, and least appreciated, it assumes that a gram of carbohydrate behaves like any other gram of carbohydrate. It does not. Fat and protein in the same meal slow gastric emptying and blunt the early rise while extending a later one, so a high-fat meal with the same carbohydrate count produces a different glucose curve — a finding documented in controlled feeding studies in type 1 diabetes and now built into some of the more advanced dosing strategies. The ratio is a linear coefficient applied to a number that is only part of what determines the response. It is useful, it is not complete, and none of this is a reason to change anything without the team who manage the treatment.

Frequently asked questions

Is the 500 rule a way to set my insulin doses?

No. It is an estimate used by clinicians as a starting point, which is then tested against real readings and adjusted, often several times. It uses one input — the total daily dose — and takes no account of time of day, activity, illness, meal composition or anything else that moves glucose. If you are counting carbohydrate, the ratio you use should be the one your diabetes team arrived at with you, not one a calculator produced from a single number.

Should the total daily dose include basal insulin?

In the form of the rule taught in most diabetes education material, yes: the total daily dose means all insulin taken in a day, basal plus every bolus. That is worth stating because using bolus insulin alone would inflate the ratio substantially and produce an estimate nobody intended. It is also one more reason the rule is only a starting point — the split between basal and bolus varies between people and regimens, and the rule collapses that structure into a single sum.

Why does my ratio seem different at breakfast?

Because insulin sensitivity is not constant through the day. Cortisol and growth hormone rise in the early morning and work against insulin, so many people are genuinely less sensitive at breakfast than later on and need a tighter ratio then. A rule that takes only a total daily dose cannot see any of that; it returns one number for twenty-four hours. Time-of-day ratios are worked out from readings, with the team, and are one of the commonest refinements made after a starting estimate.

Does a high-fat meal need the same dose as a low-fat one with the same carbohydrate?

Controlled studies in type 1 diabetes have shown it often does not behave the same way. Fat slows gastric emptying, so the glucose rise arrives later and lasts longer, and the same carbohydrate count can leave a person low early and high several hours later. Some teams address this with split or extended dosing. It is a well-recognised limitation of counting carbohydrate alone, and it is squarely a matter to work through with a diabetes team rather than to solve with a ratio.

Which is right, 500 or 450?

Neither is right in the abstract; they were fitted around different insulins and different degrees of caution. The 450 version came from the era of regular human insulin and gives a ratio 10% smaller, which asks for 11.1% more insulin for the same food. The 500 version accompanied the rapid-acting analogues. Which starting point suits a given person is a clinical decision, and both are only starting points — the real ratio is the one that emerges from testing.