Choose a medical question bank by testing the tasks you need to complete, not by relying on a broad accessibility label. Reading a question is only the beginning: you also need to select an answer, understand feedback, inspect images and resume work. A barrier anywhere in that sequence can make an otherwise useful resource unsuitable.
The checklist below is a proposed purchasing test, not an accessibility certification or a report of completed testing. No provider has been scored under this protocol. Use it on your own device and, where appropriate, with the assistive technology and support you normally use.
Begin with a learning task rather than a diagnosis label
A description such as "dyslexia-friendly" does not specify which functions work or how they have been assessed. Two people with the same diagnosis may have different practical requirements, while people with different diagnoses may encounter the same barrier.
Write down the task that matters to you: reading a long stem at a comfortable size, moving without a mouse, hearing the structure of an explanation, or understanding feedback without relying on colour. That statement is easier for a provider to investigate than a request for a general assurance.
The W3C's preliminary accessibility checks, consulted on 19 September 2026, provide a useful starting point for examining access barriers. They are explicitly an initial review, not a complete evaluation of conformance. The question-bank tasks here apply that approach to a particular purchasing decision.
Test a complete question, not just the landing page
Open a permitted sample and move through the whole sequence. Can you identify the question, its answer options and the submission control? Is it clear whether selecting an option submits it immediately? After submission, can you find the explanation and understand which response was recorded?
Try a question you do not already know. Familiarity can conceal an access problem because you may infer information that was not actually presented clearly. The test should establish whether you can obtain the content, not whether you can guess what the interface probably means.
Record the exact page, device, browser or app version and date. A screenshot can be helpful where permitted, but avoid capturing protected questions for redistribution. A short description of the barrier and steps to reproduce it may be sufficient for a support request.
Increase text size and inspect what moves
Use the display and zoom settings you would normally require. Check the stem, options, explanation, navigation and any timer. Do controls disappear, overlap or require awkward movement between parts of the same sentence?
A larger font on the question alone may not solve small labels in feedback or image controls. Similarly, a layout that looks acceptable in a screenshot may require repeated horizontal scrolling during a real task. Describe the effect on completion, not just whether the screen appears attractive.
This article does not prescribe one ideal font, colour or zoom level for all readers. The meaningful observation is whether the chosen configuration enables your task without losing content or control.
Follow the keyboard and the reading order
For keyboard use, attempt to reach each actionable element, identify the current focus and leave any opened panel. Check what happens after an answer is submitted or an explanation expands. A task can begin accessibly but become difficult when focus moves unexpectedly.
With a screen reader, examine whether the question, options, selected state and feedback are distinguishable. A sequence of unnamed buttons may expose an interface technically without making it usable. Note whether headings and controls convey meaningful structure.
Do not assume that a successful test with one browser and screen reader establishes compatibility with all combinations. Conversely, one unsuccessful setup does not identify the cause without further investigation. Report the configuration and observed behaviour precisely.
Check images and feedback separately
Medical questions may depend on visual information. Establish what the resource supplies for the intended learning task and whether an alternative description is available. A generic image label is not equivalent to an educationally useful account of the relevant findings.
Also distinguish accessibility from the assessment's purpose. Where recognising an image is part of the intended task, the examining body's arrangements need their own consideration. A commercial tool cannot decide those arrangements through an image description alone.
For feedback, cover or ignore colour temporarily and check whether correctness, warnings and progress remain understandable through text or another cue. The result should not depend on identifying one colour in a chart.
Interrupt the session and try to recover
Leave the task in a permitted way, then return. Can you locate the same question, identify the answer already submitted and resume reading the explanation? What happens if a session expires or a panel closes?
An original fictional example illustrates why this matters. A candidate can read every question comfortably but repeatedly loses their position after opening a reference panel. The problem is not text size or medical knowledge. It is session recovery, and the provider needs that specific report.
Use three outcomes in your record: completed independently, completed with a workaround, or not completed. Add the consequence of the workaround. A technically possible route that is exhausting or repeatedly loses work may still be unsuitable for sustained study.
Compare the evidence providers actually publish
As reviewed on 19 September 2026, AMBOSS publishes an accessibility statement dated June 2025, describing its approach and limitations. That is useful evidence to inspect, but it is not a substitute for testing the current task on your configuration.
This article is published by iatroX and applies the same standard to iatroX. Its September 2026 free questions offer an opportunity to inspect the learning experience before purchase. This article does not certify screen-reader compatibility or claim that the platform meets every reader's needs.
If a required task cannot be completed, another provider may be the more suitable primary bank. If the task works but one optional feature does not, decide whether that limitation affects your actual study plan. The verdict should follow the barrier and the evidence, not the brand.
Frequently asked questions
Is an accessibility statement enough to choose a question bank?
No. It is useful background, but a task-based check on your device can reveal requirements not resolved by the statement.
Does this checklist establish WCAG compliance?
No. It is a practical initial test for learners, not a comprehensive technical audit or certification.
What should I send a provider when something does not work?
Describe the task, configuration, date, steps and observed barrier without sharing unnecessary personal information or redistributing protected content. Ask for a specific response rather than a general assurance.
Try a question and inspect the complete learning experience →