The short version
- Guidelines put exercise, education and weight management first — injections are for persistent pain after those, not instead of them.
- An injection treats inflammation. If the problem is mechanical, structural, or not in the joint at all, it has nothing to work on.
- Some presentations need a different workup, not a needle — and a few of them are urgent.
- "It didn't work" is diagnostic information. A genuine non-response should change the plan, not prompt a repeat.
Almost everything written about joint injections is written by people who sell them. That produces a predictable gap: plenty on what they do, very little on when they are the wrong call.
This is that list. None of it is an argument against injections — I would not have given thousands if I thought that. It is an argument for giving them when they can actually work.
When the basics have not been tried
Both the OARSI and ESCEO guidelines for knee osteoarthritis place education, structured exercise and weight management as core treatments, with intra-articular injections positioned for pain that persists despite them.
What that actually involves is its own page. In practice the order often runs backwards. The injection is quick, it is billable, and it requires nothing of the patient in the room. Structured exercise requires a referral, several weeks, and someone actually going.
That is not an argument against injecting someone who is in genuine pain now. It is an argument against injecting instead of the thing with better long-term evidence, and then repeating it when the pain returns — which is the pattern the two-year quarterly-injection trial should make everyone cautious about.
When is the pain mechanical rather than inflammatory?
Corticosteroid suppresses inflammation. If inflammation is not what is generating the pain, there is nothing for it to act on.
The distinction is imperfect but useful, and it mostly comes down to what the pain does at rest.
| More inflammatory | More mechanical | |
|---|---|---|
| At rest | Aches, and often worse | Usually settles or disappears |
| With movement | Often eases once going | Reproduced by a specific load or motion |
| First thing in the morning | Stiff and sore, easing over the day | Fine, until the joint is loaded |
| Swelling, warmth | Frequently present | Usually absent |
| Typical cause | Synovitis, bursitis, inflammatory arthritis | Meniscal tear catching, lost cartilage, instability |
| Response to a steroid injection | Something for it to act on | Little to act on |
Injections into the second category are the most common category of disappointment I see. They sometimes produce a short benefit from the volume and the local anaesthetic, which fades in days and gets misread as a failed injection rather than a mismatched one.
The question that predicts non-response
Asking what the joint is like first thing in the morning, before doing anything, separates these two better than almost any other single question. Stiff and sore on waking, easing as the day goes on, points one way. Fine until loaded, then sharply painful, points the other.
That is pattern recognition from thousands of these rather than a validated tool. But when the answer is clearly the second, I say beforehand that this may not do much — which turns a disappointment into an expected result and a reason to look further.
When is the joint not really the problem?
A surprising share of shoulder, hip and knee pain is not coming from the joint being pointed at.
- Hip pain that is actually spine. Lumbar nerve root irritation refers to the buttock and lateral thigh convincingly. Injecting the hip does nothing for it.
- Shoulder pain that is actually neck. Cervical radiculopathy commonly presents as shoulder and upper arm pain with a normal shoulder examination.
- Knee pain that is actually hip. Classic in children and still true in adults — hip pathology refers to the knee.
- Pain around a joint rather than in it — bursitis and tendinopathy respond to different targets than the joint space, which is one of the arguments for imaging guidance.
The tell is usually a mismatch between the story and the examination: a joint that moves well and is not tender, attached to a patient in significant pain.
When it needs a different workup entirely
Some presentations are not injection candidates because something else is going on, and a few are urgent.
Not a needle — a workup
A hot, swollen, acutely painful joint with fever may be infection or crystal arthritis. Fluid should be sampled before anything is injected into it, and infection is time-critical.
Several joints involved, symmetrically, with prolonged morning stiffness suggests inflammatory arthritis, which needs diagnosis and systemic treatment rather than a joint-by-joint approach.
Pain that is unrelenting, present at night and at rest, with weight loss or a history of cancer needs imaging before injection, not after.
Significant trauma with inability to bear weight needs imaging to exclude fracture first.
When has the joint run out of runway?
In advanced arthritis, where the cartilage is largely gone and the pain is bone against bone, injections do progressively less. They can still take an edge off, and there are good reasons to use them — buying time to a scheduled surgery, or for someone who is not a surgical candidate.
What they should not be is an indefinite substitute for a conversation about joint replacement — a decision worth making on its own terms, including what recovery afterwards involves. The pattern to watch is the one covered on the frequency page: each injection lasting less than the last. That is the joint telling you something the calendar will not.
When nothing else is going to change
This is the least medical item on the list and possibly the most important.
An injection buys a window of reduced pain. If that window is used — to get through a course of therapy, to build strength, to lose weight, to change how the joint is loaded — it can be genuinely valuable. If the person returns to exactly the same activity, the same weakness and the same load, the joint arrives back where it started, having gained nothing except the time.
The measured duration of benefit is a few weeks to a few months. That is a window, not a solution. Deciding in advance what it is for is the difference between a treatment and a postponement.
Sources
- Arden NK, Perry TA, Bannuru RR, et al. Non-surgical management of knee osteoarthritis: comparison of ESCEO and OARSI 2019 guidelines. Nature Reviews Rheumatology 2021;17(1):59–66. Both guidelines position education, exercise and weight loss as core treatments. Retrieved via PubMed.
- McAlindon TE, LaValley MP, Harvey WF, et al. Effect of intra-articular triamcinolone vs saline on knee cartilage volume and pain in patients with knee osteoarthritis. JAMA 2017;317(19):1967–1975. Retrieved via PubMed.
- Jüni P, Hari R, Rutjes AWS, et al. Intra-articular corticosteroid for knee osteoarthritis. Cochrane Database of Systematic Reviews 2015;10:CD005328. Retrieved via PubMed.