Is a 30G needle the right choice for drawing liraglutide at 1 mg/mL?

Concretely: a 30G needle · liraglutide · 1 mg/mL.

I would like the axes of comparison first and the recommendation second.

I have tried the first option and it works; the question is whether the second is better rather than merely different.

What is the actual trade-off, and does it matter at the scale I am working at?

Work in the order concentration, then volume, then units, and the arithmetic stops being confusing. Concentration is milligrams per millilitre and comes from the vial contents and the diluent volume. Volume per dose is dose divided by concentration. Units on a U-100 syringe are volume in millilitres multiplied by one hundred.

Breaking it down further: if a 10 mg vial has 96.5 per cent content, you have 9.65 mg of peptide. Divide that by 2.00 mL and your concentration is 4.825 mg/mL, not 5.00 mg/mL, which is a 3.5 per cent systematic error in every dose calculation.

In practice, worked example, because the general form is easier to trust once you have seen it once. Take a 10 mg vial and add 2 mL of diluent: the concentration is 10 ÷ 2 = 5 mg/mL. A 0.5 mg dose is 0.5 ÷ 5 = 0.1 mL. On a U-100 syringe, where 1 unit = 0.01 mL, that is 0.1 ÷ 0.01 = 10 units. Change the diluent to 1 mL and the same dose becomes 5 units — same dose, half the resolution.

The content assay results from major testing services show that nominal vial claim and measured content differ by one to ten per cent, making content a driver of dose error.

Write the arithmetic on the vial label. It costs nothing and removes the step where you reconstruct it from memory.

edited 7 Jan 2025 by Dr_Idris_Coulibaly — reworded for clarity after a comment

Write the units at every step, because units errors are the failure mode that catches everyone eventually.

Dead space quantified: a fixed-needle insulin syringe holds roughly 3 to 5 µL in the hub and needle after the plunger bottoms out. A luer-lock syringe with a detachable needle holds 35 to 100 µL depending on the hub design. At 5 mg/mL that is 15 to 25 µg lost per draw on the insulin syringe and 175 to 500 µg on the luer-lock — which over ten draws is the difference between losing a rounding error and losing half a milligram.

The underlying point is that do not use the same needle to pierce the stopper and to administer. The tip is blunted by the stopper, and the hub now contains a dose you are about to lose to dead space anyway.

Worth noting: the concentration after reconstitution is not the same as the label claim, and most people do not account for the difference.

Do the arithmetic twice, ideally with someone else doing it independently.

It helps to be literal here: the common error is getting the concentration right but then misreading the syringe scale, which is why checking the barrel marking rather than your memory matters.

The concentration you actually work with is label claim times content fraction divided by actual diluent volume, which is usually not the same as the nominal concentration because content is usually not 100 per cent and you rarely measure the diluent volume to 0.1 mL precision.

The limitation is that technique reduces risk, it does not remove it, and nothing you can do outside a controlled environment makes a non-sterile preparation sterile.

If in doubt, use more diluent and accept the shorter usable window.

The arithmetic only stops being confusing once you work it through once and see that it is straightforward.

The rounding error accumulates if you round too many times — rounding concentration to 5.0, rounding the dose volume to 0.1 mL, rounding the unit reading to 10 — and the safest approach is to work the full precision and round only the final answer.

The insulin-unit standard U-100 means 100 units per millilitre, so one unit is 0.01 mL — this is the conversion that trips up more people here than any other single piece of arithmetic.

Write the arithmetic on the vial label. It costs nothing and removes the step where you reconstruct it from memory.

edited 5 Feb 2025 by halvard_ness — added a caveat about sampling