EMT Flashcards: A Study Plan for Ordered Protocol Steps
Protocol steps are ordered lists, the worst thing to put on one card. How to split a protocol into cards that test the order itself, plus a weekly review plan.
If you want an EMT flashcard study plan, here is the whole thing in four lines. Do not put a protocol on a card. Cut each protocol into cards that ask what triggers a step and what follows a step, add two or three anchor cards for the first action and the handoff, keep one scramble card per protocol as a capstone, and review daily so every joint gets scheduled on its own. The rest of this post is why that split exists, what it looks like on an actual protocol, and how to tell when yours has quietly stopped working.
I work on Nebulearn, a flashcard app, so read the tool section with the suspicion it deserves. There is a bigger limit I should put up front too: I am not an EMT. I have never run a call, I do not know what your medical director signed, and protocol wording changes from agency to agency, so every step below is a placeholder you fill in from your own protocol document. What I can talk about is the encoding problem, because it is the same one that cost me marks in second year. Ordered lists are the worst thing you can put on the front of a card, and a protocol is an ordered list.
Why a protocol on one card never actually gets tested
A card is one unit of grading. You produce an answer, you mark it right or wrong, and the scheduler moves that whole thing forward or back. That works when the answer is one fact. It falls apart the moment the answer is a seven item list, because now there are seven facts and six orderings sitting behind a single pass or fail button.
What happens in practice is that you produce four of the seven steps, recognize the other three as soon as you flip, feel like you knew it, and hit good. The card goes out to a longer interval carrying three steps you never actually retrieved. I did exactly this to myself with the Krebs cycle: one card, a paragraph on the back, a nod of recognition every review, then a blank on the exam when the question asked about one specific step out of context. The card was not lying to me. I was grading a summary.
The second problem is worse than the grading. A list card trains serial recall. You can produce step five, but only by replaying steps one through four to get there. That is a fine way to answer a written question with time to spare and a bad thing to depend on when someone interrupts you halfway. The atomic encoding rule usually gets sold as a way to make cards shorter. For a protocol it does something more specific: it gives you entry points other than the top.
The three card shapes that survive contact with a protocol
So the goal is not to shrink the protocol card. The goal is to replace it with cards that test the order directly, without a single one of them containing the whole order. Three shapes do almost all of that work.
Trigger cards ask what has to be true before a step happens. "What do you have to observe before you move to [step 4]?" The front is a condition, the back is an action. These carry the most weight, because in the protocol document the steps sit in a tidy numbered column, and the real question is never "what is number four," it is "am I at number four yet."
Joint cards ask about adjacency in one direction. "You have just completed [step 3]. What is the next action?" One step in, one step out, nothing else. You also want a small number of reverse joints, "before [step 5], what must already be done," but only at the two or three places you genuinely mix up. Reverse cards on every joint doubles the deck for very little.
Anchor cards are the fixed points: the first action, the last action before handoff, and the one step your class keeps skipping. Anchors are what let you enter the sequence in the middle. Without them the joint cards chain into a rail you can only ride from the start.
None of the three has more than one action on the back. That is the property that makes each of them gradeable, and it is why easy cards are not the problem people assume: a card that is easy to grade honestly is doing more for you than a card that is hard to grade at all.
A worked example: seven steps, fifteen cards
Take a protocol with seven numbered steps, three of them conditional, and two spots where you know from practice quizzes that you flip the order. Here is the count, and the arithmetic is the point rather than the specific numbers.
| Shape | How many | Front looks like |
|---|---|---|
| Forward joints | 6 | "You just did step n. Next action?" |
| Reverse joints | 2 | "Before step 5, what is already done?" |
| Trigger cards | 3 | "What sends you to step 4 instead of step 5?" |
| Anchors | 3 | "First action?" / "Last thing before handoff?" |
| Scramble capstone | 1 | Five steps listed out of order, back has the order |
Fifteen cards. Six forward joints because seven steps have six gaps. Reverse joints only where you actually confuse things. One trigger per conditional step. Three anchors. One capstone, the only card in the set that holds the full sequence, so you can confirm the parts assemble.
Fifteen cards at a few seconds each is a short pass. The single list card took about as long and taught you the gist. That is the trade, and it stops looking like a trade once you notice that fifteen cards get fifteen independent schedules. If you have the trigger for step four cold and keep dropping the joint between five and six, a scheduler that works per card will keep showing you the one you drop. FSRS does that by default in most modern apps, and it is wasted entirely on a deck where one card equals one protocol.
The capstone deserves a caveat. Write it as a recognition task, five steps out of order with the correct order on the back, and only add it once the joints are in. Added first, it becomes the list card with extra steps.
The failure case: a chain you can only run from the top
The split has a specific way of going wrong, and the symptom is worth knowing because the deck looks healthy the whole time.
You build forward joints for all six gaps. Every one passes. Then a practice test asks you to start in the middle and you stall. You learned the chain instead of the joints: when the card for step five came up, you silently ran one, two, three, four to reach it, and marked it correct. The app cannot see that. It shipped the card out to a long interval on a pass that was really a replay.
Two things fix it. First, watch for the replay while you review. If answering a joint card required counting up from the start, mark it wrong even though you arrived at the answer. Same discipline as refusing to count "felt familiar" as a pass. Second, break the rail on purpose. You do not need every forward joint. Leave out the two or three you have never once missed, so the deck has gaps and no unbroken path from step one to the end. A deck with holes in the chain forces the anchors to do their job.
If one joint keeps failing after a handful of honest misses, it is usually wording rather than memory. Two steps phrased with the same verb will collide forever. Rewrite one of them before you reach for a leech rule, and if the rewrite does not take, suspend it and keep the anchors on either side.
When the order is not real, and this backfires
Not every numbered list in a protocol document is ordered. Equipment checks, contraindication lists, documentation fields and most differential lists are sets that happen to be printed in a column. Adjacency cards on a set teach you something false, and that error is hard to spot later because the card looks structurally identical to a good one.
Test it before you split: if swapping two items would still be correct in the field, it is a set. Card sets by membership and discrimination instead of by sequence. One card per item, the front naming the situation and the back naming the one item, plus a discriminator card for whichever two items you keep swapping. Cloze deletion is decent for set membership inside a sentence and poor for adjacency, which is the opposite of what most people assume. The same reasoning applies to the shorthand you are drilling alongside all this, where the context matters more than the expansion.
The other place the method costs you is revisions. When your agency updates a protocol, joint cards go stale silently, because the step text still reads correctly and only the order moved. So tag every card with the protocol name and the revision date, and treat the tag as the unit you rebuild: pull the tag, rebuild the joints from the new document, keep the anchors that did not move. That is the maintenance bill for splitting, and I would rather you see it now than find it in month five.
Where each app actually wins for this
The card shapes work anywhere, so the tool question is only about which step eats your evening. Anki wins if your class already passes around a shared EMT deck and you want add-ons and a deck that outlives the course. It is free on desktop and Android, and about $25 once on iPhone. Quizlet honestly wins the Friday before a quiz when four of you need the same list in ten minutes. RemNote wins if your protocol notes are already linked together and the cards belong inside them.
What none of them fixes is the part that sent you looking for a study plan. Cutting one protocol step into a trigger, a joint and an anchor means having the protocol page open while you type, and a regional protocol set is a PDF. That is the bottleneck I built for: Nebulearn does PDF upload with a side by side view, so the protocol stays on screen while you write three cards off one step instead of tabbing back and forth losing your place. FSRS is on by default, which is what makes fifteen separate joints better than one list, and the free tier is 1,000 saved questions with 40 weekly AI credits. One thing to know before you move anything: review history does not carry across apps, so if you already have months of intervals somewhere, finish this exam there.
The weekly plan, written out
Putting the pieces on a calendar is short, because the encoding work is where the effort actually goes.
Cut one protocol per sitting, the same day you cover it in class, while you still remember which step the instructor stressed. Do not batch six protocols on a Sunday, because Sunday is when the list card comes back. Fifteen minutes of review most days, which for a handful of protocols is comfortably enough. No new cards in the last week before the exam, since every card you add is workload on the days you need for weak joints. After each practice test, audit the deck against what you actually missed: a missed question is either a joint you never carded, a trigger you never carded, or a set you wrongly turned into a sequence. If you are pulling from a course PDF rather than lecture, the PDF to cards pass is the same job with one extra decision, which is what to leave out.
A protocol is not a card. It is a handful of triggers, a handful of joints and two or three anchors, and the reason to break it apart is that each of those needs its own schedule and its own honest pass or fail. That is the whole argument, and it is what Nebulearn is built around: get the protocol PDF open beside the editor, cut one step into the cards that test it, and let FSRS decide which joint you see tomorrow.