Anki has a plausible educational rationale: it asks learners to retrieve information and revisit it over time. That is a stronger starting point than simply rereading familiar notes. However, three different claims are often collapsed into one: retrieval practice can support learning, Anki users may perform well in examinations, and Anki makes better clinicians.
Those claims require different evidence. Treating them as interchangeable leads to exaggerated recommendations and unfair dismissals. The sensible conclusion is neither that Anki is a miracle nor that flashcards are superficial by definition. It is that the tool should be judged against the learning outcome you actually need.
Evidence for retrieval practice is not automatically evidence for one app
A randomised trial by Larsen, Butler and Roediger compared repeated testing with repeated study in medical learning and found better long-term retention with testing and feedback. The published study supports the educational value of retrieval under the conditions examined.
It was not a trial proving that a particular Anki deck is superior to every alternative. Nor does it establish that any activity labelled spaced repetition produces the same benefit. The questions used, feedback provided and interval between practice and assessment all matter when interpreting a study.
For a student, the practical implication is to include genuine attempts to retrieve knowledge and to correct errors. That can happen in Anki, in written questions, in discussion or in another well-designed activity. The software is one implementation of a broader learning approach.
What Anki-specific studies show
Lu and colleagues' study examined Anki use and its association with USMLE Step 1 performance. It belongs to a body of observational work suggesting that Anki use can accompany stronger examination performance. An association is useful, but it is not the same as random assignment to Anki versus another method.
Students who choose Anki may also differ in study time, prior attainment, motivation, question-bank use and consistency. Researchers can account for some measured differences, but adjustment does not remove every possible explanation. Self-reported resource use can add further uncertainty.
Consequently, "Anki users scored better" should not be translated into "installing Anki will raise your score by this amount". It is especially unhelpful to convert a statistical association into a fixed number of cards supposedly required to purchase another examination point.
Results do not necessarily transfer between examinations
A 2023 study by Wothe and colleagues reported an association between daily Anki use and Step 1 scores, but not Step 2 scores. That finding does not prove that Anki has no value for Step 2. It does show why a broad claim about all medical examinations goes beyond what one study can support.
A later study of Step 2 CK resource use likewise illustrates the complexity of attributing outcomes to particular revision resources. Students use combinations of products, and the presence of a resource in a successful student's routine does not isolate its contribution.
The assessment itself matters. A basic-science question, a clinical management vignette and an observed consultation make different demands. A tool may be useful across all three while needing different accompanying activities and producing different measurable effects.
Retention, transfer and clinical performance are separate outcomes
Remembering the answer to a familiar prompt is one outcome. Applying the same concept to a differently worded problem is another. Performing a safe, patient-centred consultation adds communication, practical skills and judgement in context.
This distinction is not an argument against memory. Clinical reasoning needs accessible knowledge. It is an argument against using a convenient proxy as though it measured the whole task. A high review-completion rate cannot establish that a student notices the crucial feature in an unfamiliar case.
When evaluating a study or a vendor claim, identify the endpoint. Was it satisfaction, usage, immediate quiz performance, delayed retention, an external examination or observed clinical practice? A result can be genuine and still answer a narrower question than the headline implies.
What the evidence suggests for an individual study plan
Use Anki when selected facts repeatedly disappear and when the review burden is sustainable. Avoid treating it as the default response to every learning difficulty. A misunderstanding needs explanation; an unhelpful card needs rewriting; a practical skill needs practice in the relevant setting.
Keep some new questions outside your familiar materials. Record the kinds of errors you make rather than only the percentage correct. A rising score on repeated items may reflect learning, recognition or both. Unseen problems provide a useful additional check.
Periodically ask whether the next hour should go to recall, understanding or application. There is no universal allocation established by the research discussed here. Your recent performance and current assessment requirements should guide the choice.
Where iatroX fits without overstating the evidence
iatroX question banks and the Socratic Tutor offer different learning tasks alongside flashcards: answering clinical questions and discussing the reasoning behind a decision. Available simulations provide another form of rehearsal.
These product features should not be described as proof that iatroX improves examination scores or patient outcomes more than Anki. A head-to-head outcome claim would require appropriate comparative evidence. The practical proposition is narrower: students can retain useful facts in one system and deliberately practise applying them in another.
That is a more defensible basis for choosing resources than either worshipping a review streak or dismissing all memorisation. The question is whether your combined activities produce durable knowledge that remains usable when the wording, setting and task change.
The companion analysis of spaced repetition for USMLE shows how this distinction between retention and application changes a practical study plan.
Frequently asked questions
Has research proved that Anki causes higher medical examination scores?
Anki-specific studies commonly report associations rather than definitive causal effects. Retrieval practice has experimental support, but that does not establish a guaranteed score improvement from a particular app, deck or daily card count.
Does spaced repetition prove better patient care?
Not by itself. Retention, examination performance and patient outcomes are different endpoints. Evidence about one should not be presented as evidence about another without a study that actually measures the claimed outcome.
Should I stop Anki if my question-bank score is not improving?
First examine the errors. Continue targeted recall when missing facts are the problem. When the issue is interpretation or reasoning, allocate more time to explanation, unfamiliar questions and relevant supervised practice.