Right. Because by default they must have surgery if PT fails. So how do you pick your subset. Retrospectively, I presume.
Couple ways I think about this:
1) If patient has synovitis/capsular inflammation (on MRI or US), I absolutely inject into and around the joint and generally see decent lasting results (2 months +)
2) If just degenerative changes at the joint, and some bright marrow signal at distal end of the joint there is a shorter response.
3) If degenerative changes and NO increased marrow signal, then I don't expect much of a response - I usually don't bother injecting.
4) If there is distal osteolysis of the clavicle, then I don't inject since I consider that more of a fracture than arthritis.
http://radiopaedia.org/cases/post-traumatic-osteolysis-of-outer-end-of-clavicle
- You'll note a lot of this prognosticating is based on MRI and not plain film. You'll also note I'm not posting any reference (so take this all with a grain of salt) because as of my knowledge, no one has published on MRI findings correlating to injection response - doubt we'll ever see something like that either (although hundreds of these injections are done every day - you'd think someone would have made that a small resident project).
If I want to inject both the AC joint and subacromial space as targets, there are occasions where the capsule is ruptured or there is an easy path trans-AC joint to inject the medication subacromial (contrast confirms under fluoro or with color doppler flow of medication on US - probably requires another hand). That's called a "geyser" sign through the ACJ.
http://radiopaedia.org/articles/geyser-sign
For the OP, US landmarks for subacromial is fairly simple. Find the distal acromion (shadowing) and inject beneath, HOWEVER, the doesn't guarantee bursa. To get bursa, image deep to the deltoid, locate the bursa, place a needle there and distend with medication.