Anki probably carried you through the first two years. The deck was the routine. You did your reviews on the bus, before lecture, in the gaps, and the green graph kept you honest all the way to Step 1.
Then rotations started, and the deck quietly fell apart. The reviews piled into the thousands. The cards you made got worse, then stopped. By the third week of your first clerkship, opening Anki mostly produced guilt.
It is worth being clear about what actually broke, because it was not the idea. Spaced repetition still works on the wards. The part that stopped working was the card-making, and that is a workflow problem, not a science problem.
What Anki got right, and still does
The evidence behind Anki is about as settled as anything in learning. When John Dunlosky's team reviewed ten common study techniques and ranked them, the two that came out on top were practice testing and distributed practice: quizzing yourself, and spacing it over time.1 The spacing effect itself has held up across decades of studies and hundreds of experiments.2 Anki is a clean machine for doing both, and that has not changed.
So this is not an argument against Anki. If your deck still fits your life, keep it. The problem is narrower and more practical.
Why the workflow breaks on the wards
Four things happen at once when you hit clinical years.
The free time disappears, and card-making is the expensive part. Spacing and testing are cheap. Writing good cards is not. After a 13-hour day on your feet, you are not going to sit down and author clean front-and-back cards, and the half-made ones you do write are not worth reviewing.
Premade decks are built for the exam, not the patient in front of you. Zanki and AnKing are extraordinary for board facts. But the thing you most need to remember from today, the way an attending framed a hard call, a presentation that did not read like the textbook, the reason a plan changed at the bedside, is in none of them.
Clinical knowledge does not cut into clean cards. A board fact has an obvious front and back. What you learn on rotations is contextual and tied to a person and a moment. It resists being chopped into a cloze deletion, so it tends not to get captured at all.
So most people just stop. The deck goes from four hundred reviews a day to a backlog you avoid. The habit that defined preclinical study has no obvious form on the wards.
The science did not change. The fuel did.
Here is the part worth holding onto. Retrieval and spacing are still the highest-yield things you can do with your study time. The reason retrieval wins shows up even at the level of single words: people who generate an answer themselves remember it better than people who just read it, an effect robust enough to have its own name, the generation effect.3 And when students are made to recall material rather than reread it, they remember far more a week later, even though rereading feels more productive in the moment.4
Talking through a patient on the drive home is exactly that act. You are pulling the case back out of your own head, in your own words, with no prompts. That is a free-recall test on your own clinical experience. You just are not calling it studying.
What spaced repetition on rotations actually needs
If you strip the problem down, the version that survives clinical years has three requirements.
- Capture has to be almost free. If it takes more than a couple of minutes, you will not do it when you are tired, which is always.
- The cards should come from your own cases. A patient you actually cared for is already the deepest version of the material. Emotion and personal stakes are among the strongest signals the brain has for what to keep, so your own encounters lay down richer memory than a stranger's vignette.
- The scheduling has to happen for you. Nobody is going to manually resurface the case they saw three weeks ago on the exact day they are about to forget it. That is the kind of bookkeeping a tool should carry.
A few ways to keep it going
You do not need a perfect system. You need one that survives a bad day.
Keep a tiny rotations deck, and only add the one thing that surprised you. One card a day beats rebuilding AnKing and beats zero. Resist the urge to make it comprehensive.
Use the post-call window for retrieval out loud. The drive or walk home is dead time that happens to be perfect for a free-recall rep. Tell the story of the day before it fades.
Let something else make and schedule the cards. This is the loop Debrief is built around. You talk for a couple of minutes after a shift, and it pulls out the pearls and the questions and turns them into flashcards on the same kind of spaced-repetition schedule Anki uses, drawn from the patients you actually saw instead of a premade deck. Anki export is on the way, so the cards can travel with you rather than living in one more silo.
None of these replaces the science. They just lower the cost of doing it.
The one-line version
Spaced repetition is still the best study habit you can have on rotations. The fix is not a better deck. It is making capture cheap enough that you keep doing the thing that already works, on the only material that is truly yours.