Getting the drug right in a child who has to keep breathing.
A 5-year-old for MRI. You give 3.5 mg/kg propofol; she is asleep in forty seconds and the scan starts. Everyone relaxes.
Open the workbench, set up for A1 →Ce climbs for about 4½ minutes after the syringe is empty and peaks near 3.6 µg/mL — deep enough for apnoea, three minutes after everyone stopped watching. The three top-ups take Ce from 3.9 to 5.2: general anaesthesia, not sedation. The predicted time to Ce 1.0 doubles from about 18 to 37 minutes.
The child moved because the first bolus was redistributing away from a brain that had only just peaked — not because the dose was too small. Re-dosing before peak effect stacks boluses on a rising concentration. Wait for the peak before you judge.
Same child, same scan. This time the pump targets an effect-site concentration of 2.5 µg/mL.
Open the workbench, set up for A2 →One calculated bolus (about 2.8 mg/kg), plasma to about 4.6, and Ce arriving at 2.5 by four minutes with no overshoot. The rate then falls: roughly 11.7 → 10.5 → 8.9 → 7.4 mg/kg/h over forty minutes. About 8.5 mg/kg given; predicted 27 minutes to Ce 1.0.
The falling rate is the pump replacing less as the muscle compartment fills. A fixed manual rate that was right at minute five is overshooting by minute thirty. TCI is not "less drug" — it is the right amount at the right time, with a wake-up prediction thrown in.
Intranasal dexmedetomidine has taken the edge off; the cannula is in for a CT with contrast. You start plasma-targeted TCI at 1.5 and the child is still restless.
Open the workbench, set up for A3 →At five minutes on plasma target 1.5, Ce is still only about 1.1. After the step to 2.0, Ce reaches 1.4 at one minute, 1.6 at three and 1.8 at six. With an effect-site step, plasma spikes to about 2.8 and Ce is at 2.0 within three minutes.
Titration is a series of steps and waits, and the wait depends on which concentration you are targeting. Judge a plasma step at five minutes, an effect-site step at three. Impatience, not the drug, is the usual cause of oversedation in a child who was only restless.
A gastroscopy list. Deep sedation at Ce 3.0 for scope insertion. One case takes 20 minutes; the next, a therapeutic case, runs for 60.
Open the workbench, set up for A4 →20 minutes: about 6.3 mg/kg given, 29 minutes to Ce 1.0. 60 minutes: 12 mg/kg, 46 minutes. The decrement curve rises steeply in the first hour and then flattens.
Propofol's context-sensitivity is real but bounded. For short procedures the target you finish on matters more than the time you ran — step down to 2.0 for the last five minutes and watch the prediction shorten.
A 6-month-old, 7 kg, for a brain MRI. Which model do you put in the pump?
Open the workbench, set up for A5 →V1 0.41 / 0.46 / 0.23 L/kg for Kataria / Paedfusor / Eleveld; the weight-based models give twice the loading dose. Kataria and Paedfusor are flagged "outside validated range"; only Eleveld carries a maturation term. Sixty-minute totals converge (12.6, 14.3, 13.0 mg/kg) even though the starts differ two-fold.
Below one year, model choice is the biggest single decision on the pump, and no model has enough infants in it. This is where processed EEG earns its place: titrate to the brain you can see.
Multiple extractions under sedation; the list runs 90 minutes.
Open the workbench, set up for A6 →V1 0.46 L/kg on Paedfusor against 0.10 on Eleveld. Sixty-minute dose 656 mg versus 532 mg (Kataria: 902 mg). After 90 minutes, Paedfusor predicts about 60 minutes to Ce 1.0; Eleveld about 29.
Weight-proportional models scale everything by the kilograms on the scale, fat included; Eleveld scales V3 to fat-free mass and clearance allometrically. In obesity, per-kilogram dosing quietly becomes overdosing — and the models disagree by a factor of two about recovery.
A district hospital with one ordinary syringe pump and no TCI software. Can you hold a steady concentration by hand?
Open the workbench, set up for A7 →McFarlan: plasma settles at about 3.6 by 13 minutes and holds within 0.1 for the hour — a staircase drawn under the TCI curve. Bolus + fixed rate: plasma sags to 2.7 at 13 minutes (light — the child moves) and creeps back to 3.0 by the hour (deepening — the slow wake).
A stepped regimen is a TCI pump with a human doing the arithmetic; apps such as iTIVA make the steps explicit. Write the rates and the times on the syringe label — and the sixth thing you write is the time you will step down for emergence.
Induction, maintenance and emergence — and the numbers nobody has measured in your patient.
Inguinal hernia repair, LMA, caudal block. Effect-site induction at 4.0, maintenance at 3.0, wake-up planned at Ce 1.2.
Open the workbench, set up for B1 →Induction bolus about 4.6 mg/kg, plasma peaking near 7.5, Ce 4.0 at four minutes. Rate nearly 20 mg/kg/h at 5 minutes, 11.8 at 15, 9.4 at 45. About 12 mg/kg given by 45 minutes. The wake-up prediction shortens when the last minutes run at a lower target.
Small children need more per kilogram and the requirement falls continuously — no fixed rate can follow it. Emergence is planned at the thirty-minutes-to-go mark, not at skin closure.
A three-hour case. Does duration still matter with propofol?
Open the workbench, set up for B2 →About 46 minutes to Ce 1.2 after one hour; about 50 minutes after three hours. Twelve versus twenty-six mg/kg given. Moving the wake-up target shifts the curve far more than duration does.
With propofol, how deep you finish matters more than how long you ran. A step-down in the last half hour is worth more than any amount of worrying about the clock. For fentanyl this exercise would look completely different — which is why we do not infuse it.
The surgeon wants a still, spontaneously breathing child at about Ce 3.5. Two ways to get there.
Open the workbench, set up for B3 →Effect-site targeting spikes plasma to about 7.7 — more than twice the target — as the pump front-loads. Plasma steps never exceed the target, and Ce reaches about 3.1 by twelve minutes.
Mode is a clinical decision, not a pump setting. For a child who must keep breathing, step the plasma target and accept the wait; for a secured airway, let the effect site lead.
The same 60 kg body at 12, 13, 14, 15 and 16 years. Does the model think she is the same patient?
Open the workbench, set up for B4 →Paedfusor V1: 0.46 L/kg at 12, 0.34 at 14, 0.22 at 16 — the loading dose halves across four birthdays, in steps. Eleveld: 0.10 L/kg, ke₀ 0.15, time to peak 2.8 min, changing smoothly.
Paedfusor is a paediatric front-end joined to adult Marsh, and the join shows at 13–16 years. Know where your model's seams are, and watch cumulative mg/kg on every long case — propofol infusion syndrome is a dose-and-time disease.
The same child, the same target, the same drug — and one constant that decides how the pump behaves at induction.
Open the workbench, set up for B5 →ke₀ 0.26: peak effect at 4.4 minutes, plasma spiking to about 6.2 to reach Ce 3. ke₀ 0.91: peak at 2.1 minutes, plasma only 3.9. Every decrement time on the screen moves too.
ke₀ is the least certain number in paediatric TCI and the one with the largest effect on what the pump does at induction. It is the strongest argument for watching the EEG rather than trusting the violet line.