I.e. a child enters the study as a control participant and later their older sibling receives a diagnosis of Autism. Do you switch their group to "sibling" or should that participant be removed.
Conversely, what if a child enters as a "sibling" and the older sibling loses their autism diagnosis? Again, do you switch their group to "control" or should that participant be removed.
Ethically speaking, what is the best approach?
I dont have experience in this field but you cant go wrong removing them from the study. However, switching their group would depend on your research question and other parameters. e.g. the difference in having a sibling with autism and a having a sibling diagnosed with autism (presumably a different social experience for the children). For your groups did you control for age of diagnosis? If so than changing their group to "sibling" should be clean but not changing their group to "control".
You could run any analysis with and without the subjects in the new groups to show they do/dont have a pulling effect on the outcome? (like a leave-one-out cross-validation) which could support your choice to leave them in or remove them.
If there are many cases like this you could have an interesting question in a pre/post design for siblings of diagnosed autism. --this new question would need a new ethics review and consent unless you already had clearance and consent for future use of data.
I do infant sibling research. In the first case, I would witch them to the high-risk group. In the second case, if they had a solid ASD diagnosis I would keep them in high-risk, even if they "lost" the diagnosis. I would just make sure to be consistent in the approach, so if you base risk group status on a community diagnosis of ASD for at least one older sib, make sure you have that for everyone; likewise, if you confirm diagnoses in-house, bring those kids in and do an ADOS to check.