While clinical experience is certainly a reasonable place to draw evidence from, I would caution that this is generally considered to be one of the lowest forms of evidence. It is too susceptible to bias, particularly recall bias, to be the sole source of information. I would suggest that if you are to take this position, it is very important to clearly document your successes and failures with good outcome measures and well documented, consistent, and reliable methods.

Note: a lack of supporting evidence is very different from having evidence against something. This is why the debate exists at all. There is also a lack of information suggesting is doesn't work as well.

I would argue that there are plenty of options that can be provided. I think it would be prudent to make a patient aware of alternatives, but it is not necessary to run down a list of every possible treatment out there, as that would be impractical and overwhelming for a person looking to you as the expert. However, given the prevalence of US, it is equally reasonable to cover the topic with your patient, one way or another.

I would agree that this tends to require a certain level of deception on the part of the PT, which could potentially be considered unethical as well.

I'm not sure what "covered the controls" means, but it doesn't appear to be a true sham. The dosage was set to 10% lower than patient tolerance, but how could they set this if the machine wasn't on? Also, not enough info on their "prestudy" to accept their unpublished finding of the volunteers being "unable to guess accurately" which patients were receiving the true US. I'm more worried about the patient guessing anyways. I'm not totally discounting their method, but I wonder about the level of rigour in this study. This looks more like a pilot anyways as 20 patients seems a bit low given the person-to-person variability in PT treatment. The authors themselves suggest a widespread randomized controlled double blind study of US is warrented, so even they are not convinced. A single (less than perfect) RCT is not enough evidence. I would look for a systematic review, much more recent than 1986 if possible.

Many US machines are much too big to effectively be used on a desk without running you out of room. They also fail in this regard.

Placebo effect and therapeutic relationship building alone is not justification for the use of a particular modality. There are inherent risks with virtually any modality, including US, such as burns and those related to contraindications/precautions. Though "safe use" negates most of these, it is simplistic to suggest no potential for harm. Therapeutic relationship can and should be built with every aspect of PT treatment, not through use of a particular modality. Non-specific treatment effects (including placebo effect) is also present with many if not all well executed PT treatment, and again does not justify use. I would instead base my decision of whether or not to use a particular modality on whether I can see support for it having a benefit for my patient. If I am relying on non-specific effects, I would instead try to elicit this through my "actual" treatment.