The Insulin Sensitivity Factor and the 1800 Rule

What the factor is measuring

Where the carbohydrate ratio answers how much food one unit covers, the sensitivity factor answers a different question: if glucose is already above target, how far down does one unit push it? That number is what turns a reading into a correction dose. The arithmetic is a subtraction and a division — current glucose minus target glucose, divided by the factor — and it produces the units needed to close the gap.

Work one through. Take someone whose total daily dose is 50 units. The 1800 rule gives a factor of 36 mg/dL per unit. If the reading is 250 mg/dL and the target is 120, the gap is 130 mg/dL, and 130 ÷ 36 = 3.61 units. Now run the same person in SI. The 100 rule gives 2.0 mmol/L per unit. A reading of 13.9 mmol/L against a target of 6.7 leaves a gap of 7.2 mmol/L, and 7.2 ÷ 2.0 = 3.60 units. Same person, same insulin, same answer to two decimal places — because 250 mg/dL is 13.88 mmol/L and 120 mg/dL is 6.66 mmol/L, and the whole calculation has simply been divided through by 18.016 on both sides.

Why 1800 and 100 are the same rule

The derivation takes one line, and it is worth having because it removes any suspicion that two different clinical claims are being made. Glucose is C₆H₁₂O₆. Using the IUPAC conventional atomic weights, its molar mass is 6 × 12.011 + 12 × 1.008 + 6 × 15.999 = 180.156 g/mol, so 1 mmol/L is 180.156 mg/L, which is 18.0156 mg/dL. The 1800 rule returns a factor in mg/dL. To express the same factor in mmol/L, divide it by 18.016. Do that to the numerator: 1800 ÷ 18.016 = 99.91. Round it and you have 100. The two rules are one rule that has been converted, and the constant 100 is nothing more elegant than 1800 wearing SI units.

You can check the identity at every daily dose in the table above, and it holds to two decimal places throughout: 90.0 mg/dL and 5.00 mmol/L at 20 units, 36.0 and 2.00 at 50 units, 18.0 and 1.00 at 100 units. That last row is the tidiest demonstration of all — at a total daily dose of 100 units the factor is 18 mg/dL, which is one millimole per litre by definition. The same conversion is what makes 18 the famous divisor between the two glucose unit systems, and it is exactly the number covered in the blood-sugar-units article: 18 is a rounded molar mass and nothing else.

The unit split is a real hazard

Proving the two rules are equivalent does not make the unit split harmless — it makes it dangerous in a specific way. Because the two numbers describe the same person, they look interchangeable, and they are not. A factor of 36 and a factor of 2.0 are the same clinical fact expressed in systems that differ by a factor of eighteen. Enter one where the other belongs and a correction is scaled by eighteen in whichever direction the mistake runs.

The exposure is not hypothetical. People travel, change countries, change meters, read forums and guidance written for the other convention, and use apps and pumps whose unit setting is buried in a menu. Anyone reading English-language diabetes material from outside their own country is reading numbers in the other system half the time. The defence is simple and worth stating plainly: a correction factor is meaningless without its unit attached, and any number that appears without one should be treated as unusable until someone confirms which system it belongs to. Devices that let the display unit be changed should be checked after any change, because every stored factor and target moves with it.

1500 versus 1800, and where the estimate breaks

The 1500 rule is the older sibling, built around regular human insulin. Rapid-acting analogues act sooner and clear faster, and the shift from 1500 to 1800 reflects the resulting difference in how much glucose a unit is credited with moving. Some teams use an intermediate 1700. In SI each constant is simply its mg/dL numerator divided by 18.016: 1500 becomes 83, 1700 becomes 94, and 1800 becomes 100. That is worth knowing, because a reader who has seen only the mg/dL forms may not recognise an 83 as the SI face of the 1500 rule.

The estimate is at its most fragile at the two ends of the daily-dose range, and for opposite reasons. At a very low daily dose the division produces a very large factor — at 20 units the 1800 rule gives 90 mg/dL (5.0 mmol/L) per unit — which implies that a fraction of a unit does real work, a precision that a half-unit pen or a rounded pump increment cannot reliably deliver. At a very high daily dose the assumption behind the rule starts to fail: the inverse relationship between total requirement and per-unit effect is an approximation drawn from a middle range, and severe insulin resistance is not simply a scaled-up version of ordinary sensitivity. In both cases the rule is answering a question about a person it was not fitted on.

Stacking: the arithmetic that goes wrong in practice

The most consequential error in correction dosing is not a wrong factor or a slipped unit. It is time. Rapid-acting insulin analogues do not finish working when the glucose stops falling on the meter — they keep acting for several hours after injection, and a glucose reading taken an hour in shows only part of what the dose is going to do. Correct again at that point and the two doses overlap. This is called stacking, and it is the commonest route to a hypoglycaemia that arrives hours later and appears to come from nowhere.

The reason the arithmetic hides it is structural. The correction formula — current glucose minus target, divided by the factor — has no term for insulin already given. It answers as though the body were starting from rest. Modern pumps and bolus calculators exist largely to add that missing term, tracking what is often called insulin on board and subtracting it from the calculated dose. A hand calculation, or a web calculator working from three inputs, does not know it. So the honest description of any correction-factor tool, including this site's, is that it computes one line of a decision that has at least two: how far the glucose is from target, and how much insulin is still working. The second line is the one that belongs to the diabetes team, and it is why no one should be adjusting correction doses from a web page.

Frequently asked questions

Is 1800 ÷ TDD really the same as 100 ÷ TDD?

Yes, provided you keep the units straight. The first returns mg/dL per unit and the second returns mmol/L per unit, and one mmol/L of glucose is 18.016 mg/dL. Divide the numerator: 1800 ÷ 18.016 = 99.91, which prints as 100. You can verify it at any daily dose — at 50 units the factor is 36.0 mg/dL or 2.00 mmol/L, and 36.0 ÷ 18.016 = 2.00.

What happens if I mix the two unit systems up?

The correction dose is scaled by roughly eighteen, in whichever direction the confusion runs. Treating a factor of 36 as though it were in mmol/L, or a factor of 2.0 as though it were in mg/dL, does not produce a slightly wrong answer — it produces a wildly wrong one. This is why any correction factor should always be written with its unit attached, and why the display unit on a meter, app or pump should be re-checked after any change, since every stored factor and target moves with it.

Why is stacking such a common problem?

Because the correction formula has no term for it. Current glucose minus target, divided by the factor, treats every correction as though nothing had been given before. Rapid-acting insulin keeps working for hours after the reading that prompted it, so a second correction taken too soon overlaps the first, and the combined effect arrives later as a low. Bolus calculators and pumps address it by tracking active insulin and subtracting it; a hand calculation cannot. How long to wait is a question for the diabetes team, because it depends on the insulin and the person.

Should I use 1500 or 1800?

That depends on the insulin and on the person, and it is a clinical decision. Historically 1500 was fitted around regular human insulin and 1800 around the rapid-acting analogues, with some teams using 1700 in between. In mmol/L those become 83, 100 and 94 respectively. All of them are starting estimates that get tested against real readings and revised, and none is intended to be applied without a diabetes team involved.

Why does the rule work badly at very low daily doses?

Because a division by a small number produces a large factor, and a large factor implies fine dosing. At a total daily dose of 20 units the rule returns 90 mg/dL (5.0 mmol/L) per unit, so a modest gap from target implies a fraction of a unit — a resolution that half-unit pens and rounded pump increments cannot reliably deliver. That is precisely the range where children and newly diagnosed adults often sit, which is one reason paediatric and early-treatment dosing is handled with more caution and closer supervision, not with a formula.