Corrections and how to report an error

A correction is a change to something the site said that was wrong. If you could have acted differently because of what a page said, it gets corrected. Most corrections are small and specific: a number outside the range the study reported, a finding from one scar type applied to another, a citation that points at the wrong paper or has been retracted, a grade that no longer matches the literature. Missing context that leaves a true statement misleading counts too, and it is the hardest kind to catch, so report it. Typo fixes and rewording are not corrections, because the meaning did not change.

• A wrong number, dose, timeline or risk figure.
• A claim the cited source does not support.
• A citation that is dead, points at the wrong paper, or has been retracted.
• An evidence grade that no longer matches the research.
• A true statement left misleading by what is missing around it.

You do not need a medical background, and the report does not need to be diplomatic. A blunt note about a wrong number is more useful than silence. If you are a researcher and we have misrepresented your work, tell us. Those reports go to the front of the queue.

A clear factual error with a source attached is usually fixed within days. A report that requires going back to the literature takes longer, since sometimes the evidence itself has moved and the table around one line has to move with it.

Where a claim looks uncertain while we work it out, we take it down instead of leaving it up with a note. If we conclude the page was right, we say so and explain the reasoning. Either way you get an answer.

The fix goes live on the page itself, and the page's revision date updates to the day the change was made. Dates are not refreshed on pages where nothing happened, so a recent date means the page was changed or rechecked on that day.

We keep an internal record of what changed and why. A public change history is not currently published, so treat the page in front of you as the current position, not an archive of past ones.

• The revision date updates to the day of the change.
• No date refresh on a page that was not changed or rechecked.
• No public change history at this time. The page you are reading is the current position.

We are asked occasionally to soften an evidence grade, to remove a line about a side effect, or to take down a comparison a product does not come out of well. Those are not errors and will not be changed.

Evidence is the only thing that moves a grade. A manufacturer with data we have missed should send the data. Published studies are welcome from anyone, including people with an interest in the result, and they are assessed like any other source and cited with that interest visible. A change request with nothing behind it goes nowhere.

The same rule runs the other way. We will not sharpen a criticism because somebody dislikes a company, and we will not add a warning the literature does not support.