What I don't understand is why taking medications and experimenting with diet would be mutually exclusive events and why a risk/benefit analysis of dietary approaches that you consider to be "unlikely to be harmful" would not come out in favour of giving them try.
If you start meds and mess with the diet at the same time you don't know which change resulted in improvement (if it is seen.) For this same reason I resist making more than one change in medications at a time, in order to isolate the impact of each drug.
A diet that is unlikely to be harmful to a consenting adult may not be so harmless in a 3 year old. In my reading the only conclusion I could reach was that dietary changes were unlikely to result in an improvement in my daughter's disease - so while the risk of harm may have been low, so was the risk of benefit. Some people with IBD see improvement after removing certain specific foods due to difficulty with digestion etc. but we have yet to find any food or foods that produce problems for her, so we have not seen any need to mess around with her diet.
Also from a pragmatic standpoint, I'm divorced and my kids are with their father 3 days a week. He has a hard enough time keeping up with doctor-prescribed medication regimens for things like ear infections. He would never consent to a special diet based on my wishes; and even if he did agree to try it he would not be capable of carrying it out.
As for the pubmed link - case reports are not evidence, as seekonk explained.