Some medical games and gamified educational programmes have improved measured learning outcomes in particular settings. That does not establish that every daily puzzle improves examination performance, clinical reasoning or patient care.
The useful answer depends on the intervention, the learner, the comparison and the outcome. A terminology game for undergraduates, a procedural simulator and a year-long question programme for practising clinicians should not be combined into a single promise that "games work".
This article reviews selected primary studies and evidence syntheses checked on 18 October 2026. It is not a newly conducted systematic review. It is published by iatroX, whose games are discussed as product designs rather than interventions validated by the studies below.
First specify what improved
| Outcome | An example of a relevant measurement | What it does not establish alone |
|---|---|---|
| Participation | More learners return or complete activities | Knowledge acquisition |
| Immediate learning | Better performance on a post-session assessment | Retention months later |
| Delayed retention | Better answers after an interval | Transfer to a different clinical task |
| Examination performance | Improvement on a relevant assessment | Better patient outcomes |
| Clinical behaviour | A change in an observed care process | Benefit in every patient or setting |
| Patient outcome | A measured change in a defined clinical endpoint | The same effect from a different game |
These distinctions are the organising framework for this article. They prevent a registration count, a satisfaction score and a clinical endpoint from being used as interchangeable measures of educational success.
They also make a positive finding more useful. Knowing exactly what improved helps an educator decide whether the intervention addresses the problem they are trying to solve.
Gamification and a serious game are not identical interventions
A course can add points or competition to an existing activity. A serious game can instead make an educational task part of a more complete interactive system. Neither label tells the reader precisely what the learner did.
A 2023 mini-review by Xu and colleagues discusses varied uses of gamification in medical education. It is useful orientation, but it is a mini-review rather than a new pooled estimate of effectiveness.
For a particular product, ask what happens before and after a response. Is the learner retrieving information, making a decision, interpreting an image or operating a simulated patient pathway? Does the explanation address the error? Is the idea encountered again?
Those questions are more informative than counting the number of badges on the screen.
Broad review evidence: promise with substantial heterogeneity
Gentry and colleagues' 2019 systematic review included 27 randomised trials and three cluster-randomised trials, involving 3,634 participants. The interventions and outcomes varied substantially, and only two included studies concerned gamification rather than serious games.
Some comparisons favoured games for knowledge or skills, but the evidence was rated low or very low quality. Heterogeneity prevented a simple pooled answer across the interventions.
The review therefore supports taking the category seriously without treating it as a uniform treatment. It cannot identify the effect of adding one daily grouping puzzle to a doctor's existing learning routine.
It also illustrates why the comparator matters. An engaging activity may outperform a weak alternative without showing that it is preferable to a well-designed, equally demanding active-learning approach.
A later gamification review does not remove the design problem
Van Gaalen and colleagues' systematic review, published online in 2020 and in a 2021 issue, included 44 empirical studies. Assessment and challenge-related game elements were common, and some studies reported greater use of materials and improved learning.
However, many studies lacked well-defined control groups or a clear theoretical account of why the game elements should produce the outcome. The authors called for stronger longitudinal designs.
The practical reading is not that enjoyment is unimportant. It is that increased use may be an intermediate step, and the path from that use to retained understanding should be examined rather than assumed.
A leaderboard may influence participation without demonstrating which part of the experience improved learning.
Retrieval research supplies a mechanism, not a product certificate
In a 2009 randomised study by Larsen and colleagues, 40 residents completed repeated testing with feedback or repeated study of the same taught material. Testing occurred after teaching and at later intervals, with final assessment around six months afterwards.
Mean final scores were 39% after repeated testing and 26% after repeated study: a 13-percentage-point difference, not a 13% relative improvement.
This supports repeated retrieval with feedback for the studied material. It was not a trial of a modern daily game, AI tutor or iatroX subscription. Its findings should not be transferred automatically to every product that includes questions.
For a game designer, the relevant question is whether the interaction actually includes the learning process invoked in the explanation of the product.
A 2026 terminology trial measured short-term learning
A 2026 randomised trial of MedQuiz studied 60 undergraduate students in health-related programmes. It compared traditional instruction with lectures supplemented by the digital serious game.
Post-test scores favoured the game group, and the study assessed usability and learner experience. Long-term retention was not measured.
This is evidence about immediate medical-terminology learning in that population, not about independent diagnostic performance among practising doctors. The intervention also added an activity to lectures, so a reader should consider whether the comparison isolates game mechanics from additional learning exposure.
A positive short-term result remains useful when reported accurately. It does not need to be enlarged into a claim about clinical outcomes to deserve attention.
Some research has examined patient outcomes
It would be inaccurate to say that educational games have never been linked to patient-level outcomes.
A 2014 trial by Kerfoot and colleagues randomised 111 primary-care clinicians across eight hospitals over 52 weeks. A spaced-education game used 32 hypertension questions with explanations and adaptive repetition; controls received the same content in an online posting.
The analysis included 17,866 hypertensive periods among 14,336 patients. Time to the study's blood-pressure target was modestly improved, with a hazard ratio of 1.043 and a 95% confidence interval of 1.007-1.081.
This is not a 4.3-percentage-point improvement in control rates. The historical endpoint, below 140/90 mmHg, also should not be presented as a universal current treatment target.
The finding concerns a sustained, disease-specific educational programme. It does not establish the effect of a brief mixed-topic puzzle, nor identify competition as the component responsible for the result.
What to examine before accepting an educational claim
A useful appraisal begins with the learner. Medical students acquiring terminology, residents recalling taught material and practising clinicians changing management are not the same population.
Next, identify the intervention's active ingredients. A programme may combine repeated questions, feedback, reminders, additional study time, social comparison and a particular curriculum. Calling the whole package a game does not show which ingredient mattered.
Then inspect the outcome and timing. An immediate post-test can reveal acquisition but leave retention unresolved. A delayed test on familiar items may answer a different question from performance on genuinely new cases.
Finally, consider the comparison. Equal content is useful, but the experience may still differ in contact frequency, feedback or the opportunity to practise. Those differences can be the intervention's value while limiting claims about a decorative game element alone.
Where stat and clinical connections fit
stat asks for three independent clinical decisions in its initial UK-context editions. The learner commits to an option and can then inspect the explanation, alternatives and sources. The optional confidence response is separate from the points.
clinical connections asks the player to organise sixteen concepts into four groups. Group explanations and a separate optional application question provide opportunities to examine the underlying relationship.
These features make active participation and feedback available. The studies reviewed above do not demonstrate that these particular games improve examination results, prevent deskilling or change patient outcomes.
That boundary should apply consistently to competitors as well. A source-linked explanation or a university-affiliated author does not substitute for a comparative evaluation of the actual product and its intended use.
The immediate proposition can remain worthwhile: an enjoyable challenge that gives the reader something specific to understand better.
A sensible personal trial does not require a research claim
Choose one learning question and one suitable activity. Make an attempt, review the explanation and identify the distinction that mattered.
Later, try a different example that asks you to use the same idea. Ask whether you can explain it without the previous answer visible. This is a proposed self-review routine, not a validated measure of educational effectiveness.
Notice the opportunity cost as well. A game that helps you begin revision may be useful; a game that repeatedly replaces essential examination practice may not serve the same goal.
Enjoyment can be a legitimate reason to continue. It should not be used to infer improved clinical competence, but neither must every worthwhile short activity prove a patient-outcome effect before someone may enjoy it.
What an informative iatroX evaluation could look like
A proposed study would define the skill first, rather than choose the most favourable metric after collecting data.
For example, a decision-learning study could compare a specified game-and-feedback sequence with an active alternative matched for content and available study time. It could assess reasoning on new cases after an interval, with assessors unaware of allocation where feasible.
Participation, enjoyment and confidence could be recorded separately. A stronger study would also examine attrition, prior experience and whether results differ across relevant learner groups.
Clinical outcomes would require a design appropriate to actual care, with suitable governance and enough information to account for the surrounding service. A classroom quiz score should not be quietly relabelled as that evidence.
These are suggested evaluation principles, not an announcement that such an iatroX trial has been completed or is underway.
The evidence-informed conclusion
Medical games are not one intervention with one effect size. Selected studies support some benefits, including retained knowledge and, in a specific longer programme, a modest patient-level improvement. Broader reviews also identify heterogeneity and weaknesses in study design.
For a learner, select a useful task and examine the explanation. For an educator, match the format to the objective and assess the relevant outcome. For a product maker, keep design features, engagement data and demonstrated educational effects in separate columns.
That is a stronger position than either dismissing games as entertainment or presenting every enjoyable puzzle as a proven clinical intervention.
Frequently asked questions
Is greater engagement evidence that a medical game improves learning?
It is evidence about participation or experience, depending on the measure. Learning and retention need their own assessment, even when engagement may help make them possible.
Have any educational games been studied using patient outcomes?
Yes. The hypertension spaced-education trial discussed above examined time to a patient-level blood-pressure endpoint. Its result should remain attached to that particular programme and setting.
Does this research prove that iatroX games improve examination results?
No. The reviewed studies did not evaluate those iatroX products. Their findings support investigating particular educational mechanisms, not transferring a proven outcome to a different application.
Try an attempted decision followed by a source-linked stat explanation