Detailed guides to painful problems, treatments & more

STUDY: Exercise, Steroids, and Shoulder Pain That Might Be Tendinitis (Member Post)

 •  • by Paul Ingraham
Get posts in your inbox:
Weekly nuggets of pain science news and insight, usually 100-300 words, with the occasional longer post. The blog is the “director’s commentary” on the core content of a library of major articles and books about common painful problems and popular treatments. See the blog archives or updates for the whole site.
Photo of a woman with frozen shoulder, holding her shoulder, in pain.

I just keep hearing happy stories about corticosteroid injections lately. I am starting to want one myself. Almost anywhere will do, really.

I have an elderly friend who has — I am not making this up — been injected about a dozen times in various arthritic joints over the last couple years, with near perfect and long-term relief in every case. And indeed, she does seem spry! This story startled me, because I am used to thinking of steroid injections as an unpredictable and ephemeral treatment, a needle that is unlikely to move the needle for more than a few weeks at a time. Or maybe just a few days. Or not at all.

A happy steroid anecdote for a frozen shoulder case

I lean towards pessimism about steroids for the pain of frozen shoulder, which is a cranky honey badger of a pathology. But get a load of this good news report from rehab expert Greg Lehman on July 14:

Frozen shoulder update! Feeling amazing. Had a shot 4 weeks ago and I now have almost 90% full ROM. Golf doesn’t hurt and more importantly I can give the finger vigorously to drivers who blow stop signs near my kids with impunity and no pain.

Great! But will it last? Greg had already waited a few weeks to report his success, no doubt wary of declaring victory prematurely, but I made a point of checking in several weeks further on… and he’s still in good shape. And skateboarding! So it’s still great. As anyone who has had frozen shoulder knows, arm flailing is verboten with that condition.

But did he really need that shot? Could he have gotten to the same happy place with exercise?

Or maybe just a chat with a physiotherapist? (Or himself, since he’s plenty knowledgeable?)

A major new study has compared exactly those options. Thoroughly. And I will explore it thoroughly — more thoroughly than usual. Big study, big analysis.

The “GRASP” trial: Hopewell et al. on steroids vs. exercise

This post is mostly about the “GRASP” trial, by Hopewell et al. (See also physical therapist Adam Meakin’s analysis.) GRASP is one of those forced acronyms, in this case awkwardly representing Getting it Right: Addressing Shoulder Pain. What are the odds I am going to make fun of how they aren’t “getting it right” before this is over? If you’re new here, the answer is all of the odds.

Mostly a fine study, though.

Steroid injections and exercise are both used to treat pain arising from the small muscles and tendons around the shoulder, the “rotator cuff” (not the same as Greg’s frozen shoulder, although there’s some overlap). No one really knows how well either approach works.

So Hopewell et al. thoroughly compared them in the huge GRASP trial. This is a downright enormous study for this field: hundreds of people from around the United Kingdom, with a full year of follow-up. Impressive. And perhaps not surprising for the The Lancet, an important journal.

This was a “superiority” trial, which attempts to determine which of two treatments is more efficacious. It’s debatable whether this is the right kind of trial for the job, but that’s what they chose. I hope they got it right!

Rotator cuff disorders are a big deal… maybe?

From the introduction to the study:

Disorders of the rotator cuff are the most common cause [of shoulder pain], accounting for 70% of cases. Rotator cuff disorders are often associated with substantial and persistent disability and pain and approximately half of patients continue to have pain or functional limitations for up to 2 years.

That “70% of all cases” claim is important setup — shoot, it’s literally the raison d’etre for this ginormous trial — so it should probably be supported by some good data, right? Unfortunately, their citation (to Mitchell 2005) is just garbage for this purpose: it’s a fifteen-year-old topic primer for family docs that doesn’t even adequately define rotator cuff disorders, let alone credibly establish their prevalence (not even by citing another paper).

I do not dispute that there’s a lot of shoulder pain out there, and a bunch of it could be classified as “rotator cuff disorders,” but a sloppy prevalence citation like that is worrisome. Definitely not “getting it right” so far…


And what are rotator cuff “disorders” anyway?

“Rotator cuff disorders” means “shoulder pain that’s definitely maybe caused by tendinitis and not something else.”

(Yes, it really is tendinitis-with-an-i… but “tendonitis” has become acceptable via popular misuse. It’s the hardest no-win spelling challenge in my world: either way, half my readers think I can’t spell.)

According to the study, rotator cuff “disorders” include tendinitis, tendinopathy, and impingement syndrome, plus … muscle strain? Seriously? That’s silly: muscle tears shouldn’t be considered a “rotator cuff disorder” any more than a semitendinosus tear is a “hamstring disorder.”

And the other three are just different facets of the same thing.

Not getting it right again, GRASP! This is basic stuff. If you can’t produce a sensible definition of what you’re studying, good luck studying it effectively. This is a poor foundation for any trial, let alone a huge one.

GRASP is a tender tendon study

Tendinitis, tendinopathy, and impingement syndrome all overlap so much and so messily that they are essentially all the same beast. Tendinopathy versus tendinitis is not a meaningful clinical distinction — it’s shop talk about complex, debatable pathophysiology. And impingement syndrome is just a sub-type of tendinitis, mostly referring to a distinctive symptom, not an important difference in what’s actually going on.

So this is mostly a tender tendon study. But even the assumption that it’s tendons that are hurting isn’t exactly solid ground. The GRASP trial included anyone with shoulder pain without an obvious cause… therefore tendons? That’s a bit weak.

But it is a reflection of real life, where pain like this is next to impossible to definitively diagnose, and always greatly complicated by the bizarre reality that symptoms correlate amazingly poorly with whatever signs of pathology can be found. So even if you do see calcifications on a tendon, that’s not a slam dunk).

If most other candidates are ruled out, just call it a tendinitis! Amiright? I mean what else could it be? Vitamin D deficiency? Bed bugs? Little knives? Diagnosis of exclusion! Those shoulders hurt, and we have to blame something.

Study question: Do we really, though?

The point is that this was not really even a study of shoulder tendinitis, but rather a study of unexplained shoulder pain that was presumed to be tendinitis.

The exercising! Did people really exercise, and how much?

Some patients did progressive exercise, which is rehab speak for exercise that is “a little more challenging each time,” or you can dumb that down even more to “baby steps back to normal function.” They were instructed to work on that 5 times per week indefinitely on their won, but with six optional sessions with a physio over four months to check in and perhaps reinvigorate their effort.

Sustained compliance for exercises like these is generally terrible, because ain’t nobody got time for that. A self-directed exercise program is a bit of a weak sauce when you want to know how well exercise works. The idea, no doubt, was that the follow-up sessions would help people actually get it done, and the data did show that… but not strongly. And many people didn’t avail themselves of the follow-up sessions. Ain’t nobody got time for that either.

Another big group of patients was nearly identical, minus the optional follow-up sessions. In other words, they chatted with a physiotherapist for a bit, once, and then went off on their own to exercise… or not.

It’s quite unclear how much the exercisers actually exercised. The study relied on self-reporting, which is invariably overstated by patients. People like to make themselves look good. This is another flaw in the study, but not a fatal one — because, as mentioned, sloppy compliance is just normal, and so it tests real-world “effectiveness” of instead of ideal-conditions “efficacy” (how well exercise actually works with real people, versus how well it can work).

But it’s certainly worth noting that actually doing good quality progressive exercise several times per week for months is probably a different beast. Even if that beast is as fanciful as a unicorn.

Progressive exercise fails a test?!

Hopewell et al., regarding rotator cuff disorders, emphasis mine:

Progressive exercise was not superior to a best practice advice session with a physiotherapist in improving shoulder pain and function.”

Another spin on the exercise result is that it may show, to some degree, that progressive exercise isn't all it’s cracked up to be … insofar as this was actually a test of good/ideal progression (which it probably wasn't, as explained above, and yet it was also somewhat realistic and practical). This is a bit of incredible, considering how unimpeachable the principle of progression seems to be. Ever seem like science is all about disappointing us? 😜 It sure does in the pain and rehab world, where things that “makes sense” rarely seem to hold up when tested.

Progression seems so reasonable that it’s hard for me to wrap my head around the idea that it might not actually, y’know, work. (Where “works” means that it results in recovering faster/better.) But here we are. Test all the assumptions. Most ideas/claims about health prove to be wrong. The null hypothesis usually gets confirmed, in the same way that “the house always wins.”

Obviously I am not saying that this citation is the Last Word and that progressive exercise has to be thrown out. (And, for the record, that is almost never what I am saying when I cite anything. Though occasionally it is on a topic like, say, homeopathy). It’s just really interesting that this test of progressive exercise in one context had quite a clear negative result. It pisses on my bias, that’s for dang sure. I do not like this outcome! It bothers me! It makes me want to peevishly look for flaws in the paper so I can dismiss it, as one does.

But I am resisting that impulse. For now, I’m just absorbing the message that there is some evidence that clearly casts reasonable doubt on what I have believed and want to continue believe. I’m just going to sit with that for a bit and practice being comfortable with uncertainty.

The injecting!

A couple more big groups got subacromial steroid injections — that is, under the ledge of bone on top of the shoulder. One group combined steroids with exercise and multi-session follow-up, the other with the single-session flavour.

To recap, the groups were:

  1. Exercise (with multi-session follow-up)
  2. Exercise (after one session)
  3. Steroids + Exercise
  4. Steroids + Coaching

And then the researchers kept track of how everyone did for a year, measuring them with the Shoulder Pain and Disability Index (Shoulder Pain and Disability Index).

So what’s best for shoulder maybe-tendinitis?

Nothing was better than anything. Nothing was worse than anything else. Awk-ward! 🎶

“No evidence of a difference” was found for any comparison between any of the four groups. Everyone just got better on their own at about the same speed, on average, whether they exercise, got injected, or had one chat with a physiotherapist.

Another way to put that is that nothing worked — an annoyingly familiar theme in the rehab business.

And yet another way to put it is that everything worked … a little. But we don’t know, because this was not a “controlled” study — no placebo group, no non-intervention group. Quite a few people have assumed that the single-session group was a kind of control — and so, if nothing beats it, nothing works. But in reality every patient in every group got a fair amount of initial reassurance and guidance, and that matters.

So if there’s One Flaw To Rule Them All in this study, it might be that there was no non-intervention group. Unless it’s the other big flaw…

GRASPing at conclusions

Here are several take-home points about the GRASP study:

  • Our poor understanding of what’s actually wrong with these shoulders is a serious limitation. It makes the whole GRASP trial more like a REACH.
  • As just mentioned, the lack of a more neutral comparison really makes it hard to know whether steroids and exercise are equally effective… or equally ineffective.
  • Maybe “rotator cuff disorders” are generally invincible! It is a tough problem, no doubt about it.
  • Nevertheless, it would be reasonable to use this citation to support advising people not to bother with any kind of care for shoulder pain.
  • Exercise and steroid injections are two of the very most popular treatment options for shoulder. If they don’t work, it’s not like there’s a bunch of other strong candidates waiting in the wings.
  • Some people in every group did better than others, of course. The results were an average. But, as always, good luck identifying those “responders” in advance!

Effective being roughly equal, exercise still has many benefits and fewer downsides than steroids. Adam Meakins:

If you are going to use a treatment to amuse a patient whilst natural history kicks in, why not use one that has a host of positive physiological and psychological benefits rather than all your silly poking, prodding, stabbing, scrapping or sticking them with stuff.

Denouement: an example of steroids failing

The steroids helped a bunch of people in the GRASP study … temporarily. This is the fatal flaw of steroids.

I started this post with some steroid success stories. Let’s finish with a sobering failure (in a different joint):

This older patient has a forty year history of knee trauma and pain. Steroids injections had never been tried for some reason, and then finally they were. He enjoyed more or less complete relief from this extremely old pain. He was suitably impressed. He felt a little reborn, I think. Optimism was high.

And then, like Lucy pulling the football away from Charlie Brown, the relief vanished in less than a day, almost less than an hour. The pain just came back, fast, as if the injection had never happened.

This is how it often does go with steroid injections. There are excellent and lasting results for some people. For many others? Not so much — just a delicious glimpse of normal life.

Hopewell S, Keene DJ, Marian IR, et al. Progressive exercise compared with best practice advice, with or without corticosteroid injection, for the treatment of patients with rotator cuff disorders (GRASP): a multicentre, pragmatic, 2 × 2 factorial, randomised controlled trial. Lancet. 2021 07;398(10298):416–428. PubMed 34265255 ❐ PainSci Bibliography 52103 ❐


PainSci Member Login » Submit your email to unlock member content. If you can’t remember/access your registration email, please contact me. ~ Paul Ingraham, PainSci Publisher