Pediatric meal-bolus dosing pipeline — carb counting, insulin-to-carb ratio, correction factor, and active-insulin subtraction, as computed by a modern pump bolus wizard
Every insulin dosing decision in type 1 diabetes management begins with an accurate carbohydrate estimate. Overestimate the carbs and a child risks hypoglycemia; underestimate them and post-meal hyperglycemia follows. Carbohydrate counting is a learned skill blending nutrition-label literacy, portion-size visualization, and — increasingly — smartphone carb-counting apps and photo-based AI estimators.
The Nutrition Facts panel lists "Total Carbohydrate" per labeled serving — not per package. A child’s snack bag may contain "2.5 servings," so the labeled carb count must be multiplied accordingly before dosing.
Key line items: • Total Carbohydrate (g): the number used for bolus calculation • Dietary Fiber (g): for very high-fiber foods (>5g/serving), some protocols subtract half the fiber grams since it is not fully absorbed as glucose • Sugar Alcohols: similarly partially discounted in some carb-counting systems • Added Sugars: informational only — does not change the total carbohydrate figure used for dosing
For unlabeled or homemade foods, reference databases (USDA FoodData Central) or diabetes-specific carb-counting references provide gram-per-typical-portion values.
Not every meal comes with a label — restaurant meals, homemade dinners, and buffet-style school lunches require visual estimation skills:
• Hand-based portions: a fist ≈ 1 cup starch (~45g carb); a cupped palm ≈ 1/2 cup fruit (~15g carb) • Plate method: a quarter of the plate devoted to starch/grain roughly correlates to a predictable carb range once portion size is calibrated by a dietitian • Photo-estimation apps (e.g., pump-integrated carb calculators, AI food-recognition apps): a photo of the plate returns an estimated gram count, which the caregiver can adjust • Exchange lists: "1 carb exchange = 15g" simplifies mental math — a meal of 3 exchanges ≈ 45g
Studies show caregiver carb-counting error averages 10–20g even among experienced families — this uncertainty is a major source of postprandial glucose variability and is precisely why later stages (correction factor, IOB) exist to compensate for imperfect inputs.
A pediatric school lunch of chicken nuggets, a dinner roll, corn, and 4 oz apple juice totals roughly 52g carbohydrate — squarely within the typical 45–60g pediatric lunch range used throughout this simulation.
Once carbohydrate grams are known, the insulin-to-carb ratio (ICR) translates that number into an actual insulin dose. ICR is deeply individualized — determined by body weight, insulin sensitivity, age, puberty stage, and activity level — and is typically re-derived by the diabetes care team using the empirical "500 Rule."
The "500 Rule" is the standard starting-point formula for rapid-acting insulin ICR:
ICR (g carb per unit) = 500 ÷ TDD
Where TDD is the patient’s total daily insulin dose (basal + all boluses) averaged over several days. For a child with a TDD of ~42 units, ICR = 500/42 ≈ 12 — meaning 1 unit of rapid-acting insulin covers approximately 12 grams of carbohydrate.
Smaller children with lower TDD have higher ICR numbers (e.g., 1:20, more insulin-sensitive per gram), while adolescents in puberty with high insulin resistance often have lower ICR numbers (e.g., 1:6–1:8, less insulin-sensitive per gram). ICR is frequently different at breakfast (often more insulin-resistant due to dawn phenomenon/cortisol) than at lunch or dinner — many pumps allow time-block-specific ICR settings.
Once ICR is set, the meal bolus is simple division:
Meal Bolus (U) = Carb Grams ÷ ICR
For this simulation’s default 45g meal at ICR 1:12: 45 ÷ 12 = 3.75 U
Pump bolus calculators perform this instantly and round to the pump’s minimum increment (commonly 0.025–0.05 U for pediatric pumps, allowing fine dosing for small children). This meal bolus alone — before any correction — is the dose that would be needed if blood glucose were already exactly at target and there were no other insulin still active.
ICR is re-evaluated by the diabetes team every 1–3 months in growing children, since body weight, puberty, and activity level continuously shift insulin sensitivity — a ratio that fit at age 8 will typically require sharpening (a lower ICR number) by age 13.
Meal carbs are only half the dosing equation. Current blood glucose almost never sits exactly at target, so the bolus calculator applies a correction factor — the insulin sensitivity factor (ISF) — to add insulin when glucose is high or subtract (reduce) insulin when glucose is already trending low, derived via the complementary "1800 Rule" for rapid-acting analogs.
The "1800 Rule" estimates ISF for rapid-acting analog insulins (lispro, aspart, glulisine):
ISF (mg/dL drop per unit) = 1800 ÷ TDD
For a TDD of ~36 units, ISF = 1800/36 = 50 — one unit is expected to lower blood glucose by about 50 mg/dL. (The older "1500 Rule" is used for regular/short-acting insulin, which is less potent per unit on a mg/dL basis.)
Correction Bolus (U) = (Current Glucose − Target Glucose) ÷ ISF
At a current glucose of 180 mg/dL, target 120, ISF 50: (180 − 120) ÷ 50 = 1.2 U added to the meal bolus.
If glucose is below target — say 90 mg/dL — the correction becomes negative: (90−120)/50 = −0.6 U, subtracted from the meal bolus to avoid over-dosing into a low.
The pump bolus wizard sums the two components before considering active insulin:
Combined Bolus = Meal Bolus + Correction Bolus
For the running example: 3.75 U (meal) + 1.2 U (correction) = 4.95 U combined — the dose that would be given if there were zero insulin already active in the body. In practice, correction boluses are capped in pediatric settings: many clinics set a maximum single correction (e.g., ≤3 U) and require caregiver confirmation for large corrections, since children are more sensitive to dosing errors than adults on a per-kilogram basis.
ISF and ICR are mathematically linked: dividing ISF by ICR approximates how many mg/dL of glucose rise one gram of carbohydrate produces — useful for sanity-checking whether the two ratios are internally consistent during a clinic visit.
Rapid-acting insulin remains biologically active for 3–5 hours after injection, long after its peak effect. If a new bolus is calculated without accounting for insulin still working from a prior dose, the two doses "stack," delivering far more insulin than intended and driving a child into severe hypoglycemia. Every modern pump bolus calculator tracks insulin-on-board (IOB) and subtracts it automatically.
After subcutaneous injection, rapid-acting insulin follows a characteristic time-action profile:
• Onset: 10–20 minutes • Peak effect: 60–90 minutes • Duration of action (DIA): 3–5 hours (pump-configurable; commonly 4h in pediatrics)
Pumps model the declining fraction of a bolus still "active" using decay curves — some use a simplified linear model, others a bilinear (Walsh) or exponential curve that front-loads insulin activity near the peak. This simulation uses a simplified linear model:
IOB = Prior Bolus × max(0, 1 − minutes_elapsed ÷ DIA)
For a 3.2 U bolus given 95 minutes ago with a 4-hour (240 min) DIA: IOB = 3.2 × (1 − 95/240) = 3.2 × 0.604 ≈ 1.93 U still active in the bloodstream, working to lower glucose further even before any new dose is given.
The pump wizard treats IOB as insulin that has already been "promised" to the body — subtracting it from the newly calculated combined bolus so the two doses do not add together at their overlapping peaks:
Final Recommended Bolus = Meal Bolus + Correction Bolus − IOB
For the running example: 4.95 U combined − 1.93 U IOB ≈ 3.02 U recommended. Without the IOB subtraction, the child would have received 4.95 U on top of the still-active 1.93 U — a functional overdose that significantly raises the risk of a severe hypoglycemic event, particularly overnight when children cannot self-treat.
If IOB exceeds the newly calculated combined bolus, most pumps floor the recommendation at zero and flag a "no bolus recommended, monitor glucose" warning rather than displaying a negative dose.
"Bolus stacking" — repeated correction doses given too close together without accounting for IOB — is one of the most common preventable causes of severe pediatric hypoglycemia in insulin pump users, and is the single feature most credited with reducing hypoglycemic events after pump adoption.
The bolus calculator’s final output combines every prior stage into one actionable number: meal bolus plus correction bolus minus insulin-on-board. But delivery mechanics matter too — high-fat, high-protein meals digest slowly and can cause delayed glucose rises hours later, and every dose is only as good as the ratios feeding it, which are continuously refined using post-meal glucose outcomes.
A standard "normal" bolus delivers 100% of the calculated dose immediately, matched to rapidly-digesting carbohydrate. But pizza, pasta with cream sauce, or fried foods combine carbohydrate with high fat and protein content, which slows gastric emptying and produces a delayed, prolonged glucose rise that can begin 3 hours after eating and last up to 8 hours.
An extended (or "dual-wave"/"combo") bolus splits the calculated dose: • An immediate portion (e.g., 50–60%) covers the fast-digesting carbohydrate • The remainder is delivered gradually over 2–3+ hours to match the delayed glucose rise from fat and protein
Without this adjustment, a normal bolus for a high-fat pizza meal often produces an early glucose dip (too much insulin up front) followed by a late-night glucose spike (not enough insulin remaining) — a well-documented "pizza effect" familiar to pediatric endocrinology clinics.
Bolus calculators are only as accurate as the ICR, ISF, target, and DIA values programmed into them — and those values drift as a child grows. Diabetes teams close the loop by reviewing outcomes:
• 2-hour post-meal check: glucose within ~30–50 mg/dL of target suggests the ICR is well-tuned for that meal type • Persistent post-meal highs across many similar meals → ICR number is lowered (more insulin per gram) • Persistent post-meal lows → ICR number is raised (less insulin per gram) • Pre-meal-to-2h delta patterns inform whether ISF or DIA settings need adjustment independent of ICR • Continuous glucose monitor (CGM) trend arrows increasingly feed directly into modern "smart" bolus calculators, which can suggest dose adjustments based on rate-of-change, not just a single glucose value
This iterative refinement — dose, observe, adjust — is the same feedback principle used throughout automated insulin delivery (hybrid closed-loop / "artificial pancreas") systems, which continuously adjust basal and correction delivery using real-time CGM data.
A single bolus calculation is a snapshot; durable glycemic control comes from the recurring loop of counting carbs, applying ratios, delivering the dose, and reviewing the outcome — the same four-stage cycle modeled in this simulation, repeated at every meal, every day.