Should a researcher try her best to avoid using existing abbreviations (such as IEEE, WHO, DNA, ANOVA, BMI, CERN, NASA, UNESCO, OPCW, NHS, CDC ...) that are (well-)known in her fields, when creating abbreviations for new concepts (methods, substances, studies ...) in the same field?
Or is it OK if the abbreviation for the new concept is just defined where it is used (e.g., in a publication)?
Yes, you should avoid using well-established acronyms to mean something else. I would especially avoid those like the ones you present as examples that are likely more recognizable as the acronym than what the acronym stands for: they are effectively words by themselves with a specific meaning.
Yes. But then again, no. Depends. If you can easily avoid it, sure - avoid it. It will only cause confusion in the long run. But this confusion is dependent of the previous acronym being relevant (as opposed to known) in that specific field. So if you are creating a new modular iterating algorithm (stupid example, but you catch my drift), and your last name is Brown, it is OK to call it the Brown Modular Iterator (BMI). No one, in context, will think this is the Body Mass Index.
A slightly different example of when it is OK (not, mind you, optimal) to use an existing acronym which can actually cause confusion, is when there are specific naming conventions. This is how we have the American Sociological Association (ASA), the American Statistical Association (ASA), and the American Society of Anesthesiologists (ASA) - with most societies holding the convention of country_societyname_association, and medical societies to the country_societyof_societyname.
