The IOB Guide for T1D — Part 1

The Insulin On Board–Physiology Mismatch

Device-displayed IOB is a mathematical estimate. Physiological insulin exposure is the biological reality. These are not the same thing — and the gap between them explains most of the situations that feel random, frustrating, or contradictory in T1D.

How rapid-acting insulin actually behaves

Rapid-acting insulin analogues — aspart, lispro, glulisine, faster aspart, ultra-rapid lispro — do not have a single fixed duration. In broad terms, they begin working within approximately 10 to 20 minutes, often peak around 60 to 120 minutes, and continue exerting biological effect for approximately 4 to 8 hours.

But here is the point most IOB explanations miss.

A small bolus behaves differently from a large bolus. The tail stretches as the dose increases. This matters clinically because the boluses that create the most problems — post-meal highs, exercise hypos, correction stacking — are often the larger ones.

One of the clearest ways to think about insulin dose size is units per kilogram (U/kg), which normalises dose relative to body weight and better reflects the physiological exposure created by a bolus.

Dose and duration: what the numbers look like in practice

Dose-dependent insulin action by U/kg: how larger boluses produce a longer insulin tail

You might assume this does not apply to the newer ultra-rapid insulins such as Lyumjev. However, the prescribing information itself shows otherwise (thanks to Joseph Henske, Professor of Medicine, University of Arkansas, for highlighting this). Pharmacodynamic studies of ultra-rapid lispro demonstrate faster early insulin exposure, but still show a prolonged tail of biological activity, particularly at larger doses. Faster onset does not eliminate the dose-dependent extension of insulin exposure.

Lyumjev pharmacodynamic data showing prolonged tail at larger doses

The table below shows how the same rapid-acting insulin dose expressed as units per kilogram (U/kg) translates into actual units across different body weights, alongside a conceptual estimate of how long meaningful biological activity may persist.

Note on these estimates

These durations are rough estimates, not fixed pharmacokinetic limits. Real insulin action varies between individuals and is influenced by injection site, insulin sensitivity, activity, and stacking. As insulin dose per kilogram increases, both peak timing and the duration of the insulin tail tend to extend.

Body Weight	0.05 U/kg	Peak / Tail	0.1 U/kg	Peak / Tail	0.2 U/kg	Peak / Tail	0.3 U/kg	Peak / Tail
25 kg	1.25 U	1 hr / 4 hr	2.5 U	1.5 hr / 6 hr	5 U	2 hr / 7 hr	7.5 U	2.5 hr / 8 hr
50 kg	2.5 U	1 hr / 4 hr	5 U	1.5 hr / 6 hr	10 U	2 hr / 7 hr	15 U	2.5 hr / 8 hr
75 kg	3.75 U	1 hr / 4 hr	7.5 U	1.5 hr / 6 hr	15 U	2 hr / 7 hr	22.5 U	2.5 hr / 8 hr
100 kg	5 U	1 hr / 4 hr	10 U	1.5 hr / 6 hr	20 U	2 hr / 7 hr	30 U	2.5 hr / 8 hr

Peak action is not disappearance

A common mental model is: “insulin does its main work after 2 hours.” Peak action does not represent the end of insulin activity. Reduced effect does not mean no effect.

This matters because a lot of T1D decision-making happens precisely in the window where insulin is fading but still meaningful:

the 2 to 4 hour window after a meal
the “I’m still high, so I’ll correct again” window
the “I’ve got time for a quick workout” window

If you treat this window as insulin-free because a display says 0.0, you will repeatedly be surprised by outcomes that are actually predictable once you understand exposure.

Circulating insulin is what drives tissue effects

Physiologically, what matters is insulin exposure to tissues — muscle, fat, liver — which is driven by absorption into circulation and clearance over time. That is why two people can take the same dose and experience different outcomes:

absorption variability (site, temperature, lipohypertrophy)
clearance differences (especially the kidneys)
changes in sensitivity (sleep, stress, hormones, recent activity)

The key distinction

Device IOB is a mathematical estimate. Your physiology is a dynamic biological system. These are not the same thing.

Portal vs peripheral: the gradient is reversed in T1D

In people without diabetes, insulin is secreted into the portal vein and first passes through the liver. The liver sees high insulin concentrations; peripheral tissues (muscle and fat) are exposed to lower concentrations.

In T1D, insulin is delivered into subcutaneous tissue and enters the peripheral circulation first. That reverses the natural gradient.

Portal vs peripheral insulin delivery: the gradient reversal in T1D and its consequences for muscle, fat, and exercise

Physiological feature	Without Type 1 Diabetes	With Type 1 Diabetes	Practical consequence	Implications for exercise, fat storage, and glucose control
1. Where insulin is delivered	Insulin secreted from the pancreas into the portal vein, reaching the liver first.	Insulin delivered into subcutaneous tissue, absorbed slowly into the systemic bloodstream.	The liver receives insulin immediately in people without diabetes, but much later in T1D.	Exercise risk depends more on circulating systemic insulin in T1D.
2. Initial handling of meal glucose	The liver captures and stores a large proportion of incoming glucose from the gut.	Because portal insulin is low, the liver captures less of the incoming glucose load.	More glucose escapes into the systemic circulation after meals.	Higher circulating glucose combined with circulating insulin increases glucose swings during activity.
3. Glucose entering the bloodstream	Only a relatively small amount of glucose leaves the liver and enters the bloodstream.	A larger portion of glucose reaches the bloodstream quickly, producing the familiar post-meal spike.	Even with a perfect carbohydrate ratio, glucose may rise temporarily after eating.	Higher glucose exposure before exercise can create unpredictable behaviour depending on IOB.
4. Systemic insulin levels	Most insulin acts locally in the liver; systemic circulating insulin levels remain relatively low.	Insulin circulates widely in the bloodstream because it must travel from subcutaneous tissue to all organs.	Peripheral tissues (muscle and fat) see higher insulin exposure.	Higher systemic insulin increases exercise-related hypoglycaemia risk within several hours of a bolus.
5. Glucose storage in muscle and fat	Glucose storage in muscle and adipose tissue is modest because hepatic buffering controls glucose entry.	Larger amounts of glucose are stored in muscle and fat due to higher circulating insulin and glucose.	Greater metabolic exposure to insulin in peripheral tissues.	Over time, this can increase fat storage risk and amplify exercise glucose drops.

Simply put:

The liver is under-insulinised, so glucose output is not suppressed, especially post-meal and during stress.
Muscle and fat are over-insulinised, and when combined with activity, insulin exposure can feel far stronger than expected, even 3 to 4 hours after insulin was administered.

This reversal of insulin exposure between the portal vein and peripheral tissues sits at the heart of the IOB problem. In T1D, insulin is delivered in the opposite direction to how human physiology was designed to use it. This physiological reality is precisely why exercise responses in Type 1 diabetes depend so strongly on the insulin exposure present at the start of activity.

Exercise supercharges insulin exposure

Exercise is often taught as if it simply uses glucose. In T1D, it is more accurate to think of exercise as a multiplier on insulin effects. During physical activity, several things happen at once:

Muscle blood flow increases, delivering insulin and glucose to working tissue faster.
GLUT4 transporters increase at the muscle surface (insulin-dependent and insulin-independent pathways).
Insulin sensitivity rises, sometimes sharply.
Insulin clearance by the kidneys is reduced, making more insulin available to muscle cells — insulin effectively becomes stronger.

How exercise amplifies insulin exposure: increased blood flow, GLUT4 upregulation, insulin sensitivity, and reduced renal clearance

This is where the difference between displayed IOB and physiological insulin exposure becomes critically important. A device may display little or no IOB while circulating insulin remains sufficient to amplify exercise-driven glucose uptake.

This is why two exercise sessions that look identical on paper can behave completely differently:

Morning fasted activity with no bolus insulin often leads to glucose drifting up.
The same activity in the afternoon 2 to 4 hours after a bolus can drop glucose fast.

The difference is not the exercise itself. The key variable is the insulin exposure entering the session, shaped by the cumulative bolus and correction doses earlier in the day. This is also why substantial bolus reductions of 25 to 75% are typically required when exercising within two hours of a meal bolus, because the insulin exposure is still near peak.

A familiar scenario: 0.0 IOB and still dropping

Consider this sequence:

12:00 — lunch bolus delivered
14:00 — device displays 0.0 IOB
14:15 — exercise begins
14:35 — glucose falls faster than expected

Many people interpret this as exercise being unpredictable. Often the opposite is true. The physiology is predictable. The display is not aligned to the physiology — especially when device settings are optimised for correction behaviour rather than insulin exposure.

Biology says: rapid-acting insulin commonly exerts meaningful biological effects for approximately 4 to 6 hours, with duration influenced by dose size and individual physiology. Most device models offer an AIT/DIA of approximately 2 to 8 hours, often defaulting to 3 to 4 hours, and most diabetes teams and individuals shorten this to 2 hours. The reason for the mismatch is that devices are not modelling pharmacology alone. They are balancing safety and correction behaviour using a simplified model that often groups meal and correction insulin together.

Key principles from Part 1

Insulin duration depends on dose per kilogram.
Insulin still has meaningful biological effect well after peak.
Device IOB displays may significantly under-represent this exposure.

If insulin behaves this way biologically, the next question becomes: how do diabetes devices attempt to model it? Part 2 explains the different mathematical models, why IOB often under-represents physiological exposure, and why that under-representation is frequently done on purpose.

Continue the guide

← Back to hub
Part 2: Different Models →

Back to The IOB Guide for T1D