What happens when you flag a question
Every question bank contains errors. Thousands of questions, human authors, moving guidelines: the only banks without wrong answers are the ones nobody is checking. So the honest measure of a bank is not whether errors exist but what happens in the minutes, days, and versions after someone spots one. Here is the full path in MedQVault.
The flag
Every question carries a flag control, in the player and in review. Flagging takes a few seconds, is anonymous to other users, and never interrupts your session; the question simply goes into a triage queue with your comment and the context attached.
Triage
A clinician reviews the queue. Most flags resolve into one of three outcomes:
- The question stands. Often the explanation already answers the objection, which usually means the explanation needed to make its case earlier or more plainly, and that is a fix in itself.
- The question needs work. A dated reference, a clumsy stem, a distractor that has drifted from fair to unfair. It gets revised.
- The key is wrong. The rarest outcome and the most serious one, and it is handled by the system rather than by discretion.
Auto-suspension on a mis-key
A confirmed mis-key does not wait politely in a queue while more candidates answer it. The question is suspended automatically: it stops being served to anyone, immediately, and stays out of circulation until a corrected version ships. A question teaching the wrong answer is doing damage every hour it remains live, so removal is not a judgement call.
Versioning, and why your history is safe
Fixes ship as a new version of the question, with the old version preserved. This is the part that protects you. Your response history stays attached to the exact version you actually answered, so a question can never silently change under your feet and rewrite what your past answers mean. If you answered version one, your statistics say so forever.
This matters more than it first appears, because statistics are how the bank polices itself. Response-level data accumulates into item statistics: a question most strong candidates get wrong, or a distractor nobody ever picks, surfaces in review whether or not a human flags it. Corrupt the response log with silent edits and every one of those signals becomes noise. Keeping the log immutable is what lets flags, statistics, and versioning reinforce each other instead of undermining each other.
Why we bother telling you
Because trust is the product. A bank that hides its error handling is asking you to assume it has none. Ours is built so that the worst case, a wrong key reaching candidates, has a bounded lifespan and a permanent audit trail. The quality bar that questions have to clear in the first place is covered in how MedQVault questions are built; this post is the safety net underneath it.
See the questions themselves in the free samples. And if you spot an error there, flag it. That is the system working.