Back to blog

How to Practice Dosage Calculations (Flashcards Only Do Half)

Dosage calculations are a procedure, not a fact. Card the setup and the unit conversions, do the arithmetic on paper, and stop guessing which half failed.

Hello there. If you are looking for how to practice dosage calculations, the short version is that you practice them on paper, with a pencil, one problem at a time, and you use flashcards for the layer underneath. The cards hold three things: which relationship applies, which conversion factor you need, and what the units have to cancel down to. The arithmetic itself does not belong on a card.

A little about me first, because credibility matters here and mine has limits. My name is Daniel, I am a third year engineering science student at the University of Toronto, and I work on Nebulearn, which is a conflict of interest you should know about before I say anything about apps. I am not a nurse, a pharmacist, or a clinician of any kind. What I can speak to is the math: three years of unit-cancelling problems in thermo, circuits, and fluids, where writing kg when you meant g costs you the whole question.

Most decks fail at this because they put the whole problem on one card. Front: a scenario with numbers. Back: a number. That is a practice problem with a scheduler bolted to it, and when you miss it you cannot tell whether you picked the wrong relationship, botched a conversion, or dropped a decimal. One card, three possible failures, no diagnostic.

Break the calculation into the four things it actually tests

Before you can decide what goes on a card, you have to see that a dosage calculation is not one skill. Do one slowly and it comes apart into four moves.

First, recognition. You read the situation and decide what kind of problem it is: a volume to draw up from a stock concentration, a dose scaled to body weight, a flow rate over a number of hours, drops per minute for a given set. Get this wrong and every step after it is wasted.

Second, unit alignment. The order is in one unit, the thing in front of you is labelled in another, the weight is in pounds when the formula wants kilograms.

Third, setup. You arrange the numbers so the units you do not want cancel and the unit you do want survives. mg divided by mg/mL leaves mL. If that cancellation does not work out, the number you compute is meaningless no matter how careful your long division was.

Fourth, execution: the arithmetic, plus the sanity check. Is this a plausible size, or did I just produce something ten times too big?

Moves one through three are recall and card shaped. Move four is a motor skill you build with a pencil and a stack of problems, and it is the one everybody tries to cram onto the card anyway.

Why a card that makes you do arithmetic goes bad

That split is not a style preference. An arithmetic card degrades in a review queue for three mechanical reasons.

The first is grading ambiguity. You set the problem up perfectly, cancel the units correctly, then slip a decimal. Now what do you press? Mark it wrong and you have told the scheduler you do not understand a setup you clearly do. Mark it right and you have started grading yourself on "close enough," which is the one habit this subject cannot tolerate. Coding cards have the same problem, which I wrote about in the syntax versus logic gap: once a card can fail for a trivial reason and a serious reason at once, your ratings stop meaning anything.

The second is time. A review queue works because each card is a few seconds long, which is why average review time per card is measured in seconds. Give a card a three-step calculation and it becomes a ninety second card with a pencil hunt attached. Twenty of those is a problem set, not a review session, and you will quietly stop opening the app on days you do not have a desk.

The third is the worst. The numbers on the card never change. Do that card six times and you are not calculating any more, you are remembering the answer rather than deriving it. It looks like mastery because the card comes up green, and then a quiz hands you the same relationship with different numbers and the fluency evaporates. That is one of the flashcard mistakes students make that hits procedures hardest: a procedure has a legitimate answer you can memorize instead of the method.

The setup cards that replace one calculation card

So what does the deck hold instead? Take a generic example: a stock solution labelled 250 mg per 5 mL, and an order for 400 mg. On paper that is two lines. The stock is 50 mg per mL, so 400 divided by 50 is 8 mL. I want that arithmetic to happen on paper every time, with fresh numbers.

What I want on cards is everything that made those two lines possible: three or four prompts, in the spirit of the atomic encoding rule.

Card familyFrontBack
RecognitionYou are given an amount ordered and a concentration on hand. What are you solving for?A volume. This is a volume-to-give problem, not a rate problem.
SetupIn a volume-to-give problem, which quantity goes on top and which underneath?Amount ordered on top, concentration on hand underneath.
Unit targetWhat cancels in that setup, and what survives?mg over mg/mL cancels the mass and leaves mL.
MagnitudeThe order is larger than the amount held in one full unit of stock. What should the answer look like?More than one full unit of volume. Smaller means the division went the wrong way.

None of those asks me to divide anything. Each is answerable in about four seconds, out loud, on a streetcar. And when I fail one I know what I failed: the recognition card means I am not reading problems carefully, the unit target card means my dimensional analysis is shaky.

The magnitude card is the one people skip and the one I would keep if I could keep only one. Decimal errors are the whole ballgame here, and a decimal error is a magnitude error, so carding the expected size gives you a check that runs independently of the computation.

This is where the page splits from my earlier piece on encoding math formulas. That one was about formulas you have to reproduce from memory, and carding what each variable does instead of the shape of the symbols. Here the formula is trivial and nobody forgets it. What fails is choosing the relationship, and the conversion that happens before you touch the formula.

Unit conversion deserves its own card family

Conversions are the half people bolt onto the end of a calculation card, then wonder why they keep missing questions they understood.

They are pure recall, the one part of this topic that is genuinely flashcard shaped, and they deserve their own tag. A short list covers most of it: 1 g is 1,000 mg, 1 mg is 1,000 mcg, 1 L is 1,000 mL, 1 hour is 60 minutes, 1 kg is about 2.2 lb.

Three rules for how to card them.

Card the direction separately from the factor. Knowing the factor between mg and mcg is 1,000 is not the same retrieval as knowing whether you multiply or divide going from mcg to mg. The factor is almost never what fails. The direction is. So the front is "going from mcg to mg, multiply or divide?" rather than "how many mcg in a mg?", and a second card runs it the other way, because a card you have only ever reviewed left to right does not transfer.

Card the awkward numbers, not the clean ones. 1 mg to 1,000 mcg is tidy on both sides. What breaks people is 0.5 mg becoming 500 mcg, or 750 mcg becoming 0.75 mg, with a leading zero to hold.

Card the pound-to-kilogram step as its own prompt, because weight-based dosing is where two of the four moves stack. Say the order is 6 mg for every kg and the weight you are handed is 154 lb. Convert first: 154 divided by 2.2 is 70 kg. Then scale: 70 times 6 is 420 mg. Then, and only then, the volume. Three steps, and only the first is a fact. A card for "the formula wants kg and I have lb, what comes first?" makes that step automatic.

Rate problems behave the same way. 1,000 mL over 8 hours is 125 mL per hour, and that division is paper work. But "what do I multiply by to get from mL per hour to drops per minute, given a drop factor" is recall, and it earns a card because the 60 in the denominator is what people forget. Run it: 125 mL per hour at 15 drops per mL is 1,875 drops per hour, or about 31 drops per minute. The numbers are disposable. The structure is not.

What to do when the cards pass and the quiz doesn't

Every method has a failure mode, and this one has a common one, so here is how to read it.

You get every setup card green for two weeks and lose marks on a practice quiz anyway. That almost always means the deck has no real recognition layer: you built cards for the problem types you already knew to look for, and the quiz handed you a stem that did not announce itself. The fix is not more cards from the textbook, it is mining your own wrong answers. Every time a practice problem catches you out, the card you write is the recognition card, phrased with the words from the stem that misled you. That is the post-exam review habit applied to practice sets instead of exams.

The opposite failure is one setup card you keep missing until it is a leech. Do not keep drilling it. A setup card that will not stick is rarely a memory problem, it is a card standing in for a concept you never built, so go back to a worked example, re-derive the relationship by hand until it makes sense, then rewrite the card in your own words. My leech protocol applies here like anywhere else: after enough failures, the card is the problem.

And if paper practice keeps failing at the arithmetic and only the arithmetic, stop making cards for a bit. That is ten minutes a day of long division and decimal drills, which no flashcard app will fix.

Where a flashcard app helps, and where it gets in the way

The practical question is what holds the setup deck. I have an obvious bias, so here are the cases where the answer is not us.

If you are at a placement with no signal and a spare five minutes, use Anki. It is local-first and genuinely works with the network off. Nebulearn does not: we show a connection problem banner when the browser drops, and there is no offline mode. That is a real gap, and this is a subject people revise in corridors. If your cohort already maintains a big shared deck, stay in it. And twelve conversion factors you and a classmate want to drill before a Friday quiz is a Quizlet set, honestly, because it will be dead by Monday.

Where I think we are the better answer is the part of this page that is actual work: getting the setup layer out of a question bank and into cards. Practice problems mostly live in a browser tab, and the useful card is a reworded fragment of a stem plus the unit target, not the whole question. Nebulearn's Chrome extension snips anything on the page straight into a card with a keybind, so the recognition card gets written in the two seconds after a problem catches you out. The editor has a LaTeX button for the ratio setups, and reviews run on FSRS by default, so conversion cards you know cold drift out to long intervals while the direction cards you keep reversing stay close. The free tier is 1,000 saved questions and 40 weekly AI credits, more than a conversion deck and a setup deck need.

My notes on spaced repetition in nursing school and the mechanism versus dose split cover the neighbouring decks on the same principle: the card holds what you must recall cold, and the work you must perform stays work. Setup, direction, and expected magnitude go in the deck. The arithmetic stays with the pencil.

dosage-calculationsflashcardsunit-conversionstudy-tips