What does splitting a weekly dose actually cost in vial entries and compounded measurement error?

Accepted answer

Costed in order: punctures rise 80 %, measurement error roughly doubles because reading bias is systematic rather than random, dead space is either trivial or ruinous depending on one hardware choice, and every one of these costs gets worse as the volumes get smaller.

1. Punctures and contamination probability

Count the entries over the vial's in-use life. Reconstitution is one entry regardless.

1. Weekly: 1 reconstitution + 4 draws = 5 punctures over 28 days.
2. Split: 1 reconstitution + 8 draws = 9 punctures over the same 28 days.
3. Increase: (9 − 5)/5 = 80 %.

Translating that into contamination probability requires a per-entry probability p that nobody knows for a domestic setting, but the structure is informative even so. Probability of at least one contamination event over n independent entries is 1 − (1 − p)^n. For small p that is very close to n·p, so risk is essentially linear in puncture count. If p were 1 in 1,000, five entries give 0.50 % and nine give 0.90 %. If p were 1 in 100, five give 4.9 % and nine give 8.6 %.

You cannot pin the absolute number, but you can be confident of the ratio: splitting buys 1.8 times the cumulative contamination exposure per vial, whatever the per-entry risk is. It also adds a second mechanical consideration, which is that stopper integrity degrades with puncture count and coring risk accumulates — a closure that has been entered nine times is not the closure it was after two.

The in-use window does not extend to compensate. A 28-day in-use limit is a limit on elapsed time, so splitting compresses more entries into the same window rather than spreading them over a longer one.

2. Measurement error, and why it doubles rather than growing by 41 %

This is the part people get wrong, and the answer depends on whether your reading error is random or systematic.

On a U-100 syringe with 1-unit graduations, careful reading resolves to about half a graduation, so ±5 µL. Now:

If the error were purely random, two independent draws would combine in quadrature:

1. Single 480 µL draw: ±5 / 480 = ±1.04 %.
2. Two 240 µL draws: total error = 5 × √2 = ±7.07 µL on 480 µL = ±1.47 %.
3. So a 41 % increase in the weekly error.

If the error is systematic — and it mostly is — the two errors add rather than combining in quadrature:

1. Two 240 µL draws with the same bias: 2 × 5 = ±10 µL on 480 µL = ±2.08 %.
2. Exactly double the single-draw error.

Reading error in this task is dominated by systematic components: which feature of the plunger seal you align to, the direction you habitually view from and hence your parallax sign, and whether you consistently read at the meniscus or the seal edge. These do not resample themselves between draws. They are the same person, the same habit, twice. So the doubling case is the realistic one, and anyone estimating this as a 41 % increase is assuming a randomness their own hands do not provide.

3. Dead space: one hardware choice decides everything

Splitting doubles the number of draws, and dead space is paid per draw, so the dead-space loss doubles. Whether that matters is entirely a function of which syringe:

So on a fixed-needle syringe the extra dead space from splitting costs 0.04 mg of a 10 mg vial, which is 0.4 % and beneath the precision of anything else in the calculation. On a luer configuration it costs 1.68 mg, or nearly 17 % of the vial — and worse, the totals no longer fit. Check that:

1. Luer, weekly: each draw removes 480 + 84 = 564 µL. From 2,000 µL that is 2000/564 = 3.5, so three doses, not four.
2. Luer, split: each draw removes 240 + 84 = 324 µL. 2000/324 = 6.2, so six half-doses = three whole doses.
3. Fixed-needle, weekly: 480 + 2 = 482 µL; 2000/482 = 4.1, so four doses.
4. Fixed-needle, split: 240 + 2 = 242 µL; 2000/242 = 8.2, so eight half-doses = four doses.

The syringe architecture costs you a whole dose per vial and the split costs nothing, in both cases. Which means the honest ranking of interventions is: fix the syringe first, and then splitting is nearly free on this axis.

4. The bottom of the ladder is where it all breaks

Every cost above scales badly as the volume falls, because measurement error and dead space are absolute quantities while the dose is not. Work the 0.25 mg initiation rung at the same 5.0 mg/mL:

1. Whole dose: 0.25/5.0 = 50 µL = 5 units. Reading error ±5 µL = ±10 %.
2. Split: 25 µL = 2.5 units. Reading error ±5 µL = ±20 % per draw, and with a systematic bias, ±20 % on the weekly total.
3. On a syringe with half-unit graduations you resolve to ±2.5 µL, so those become ±5 % and ±10 % respectively. Better, and still poor.
4. Dead space on a luer configuration at 25 µL draws: 84/(84+25) = 77 % of what leaves the vial is discarded.

So the general result is that splitting is cheapest exactly where it is least argued for — at the top of the ladder with large volumes and a fixed-needle syringe — and most expensive at the bottom of the ladder where people reach for it because of tolerability. A ±20 % weekly dose uncertainty on an initiation rung is a larger perturbation than any curve-flattening benefit on offer, and it is the wrong kind of error: unmeasured, unnoticed and one-directional if your reading habit is biased.

One thing that mitigates most of this: choose the reconstitution concentration so that the dose you intend to draw lands on a readable whole number of graduations, and if a split is contemplated, so that the half-dose does too. That is a decision made once, at reconstitution, and it costs nothing.

edited 8 Sept 2025 by charge_state_3 — removed a claim I could not source