Starting a few decades ago, we found that "the vast majority of ulcers are caused by H pylori infection" (quote from http://www.ulcerresourcecenter.com/ulcers-and-milk ), and can be treated with drug therapy.
I point this out as an example of how even if there is a reasonable mechanistic model of how a therapy might work, which is widely believed to be true, that doesn't mean it is true. The Wikipedia link I posted gives reference to attempts to verify if NLP is a useful treatment, and concluded that it isn't.
As for "pattern interrupts and overwriting old anchors is heavily used in behavioral therapy", if you search for '"pattern interrupts" "behavioral therapy"' on Google there are very few hits. Duck Duck Go has even fewer. This suggests that "pattern interrupts" is not a term used in behavioral therapy. If you further look at http://www.ncbi.nlm.nih.gov/gquery/?term=%22pattern+interrup... you'll see that "pattern interrupt" is not commonly used, with only 214 matches in PubMed. A spot check shows only a handful used in the sense you mean.
Therefore, the term "pattern interrupts" is likely not the term used in behavioral therapy. You are, I believe, making a statement about a homologous treatment, perhaps operant conditioning? But operant conditioning, and its application, predate NLP, so I don't think it's reasonable to attribute its effectiveness to NLP, nor would it be novel in NLP if Skinner discussed it in the 1950s.
("Operant conditioning" has 18356 matches in PubMed, and the first 7 papers all use it in the cognitive psychology sense.)
