The threshold question: when does injection become the right next step?

An injection becomes a realistic option when first-line care has been properly tried and has not been enough — not simply because a scan shows osteoarthritis.

First-line care means a genuine, sustained attempt at weight management, structured exercise, physiotherapy, and oral pain relief such as NSAIDs or analgesics. More than one in ten people with knee OA receives an intra-articular injection each year, typically within the first few months of diagnosis, yet clinical guidelines consistently position injections as a second step — initiated after those conservative measures have run their course.

The patient profile that fits this stage broadly looks like this: moderate-to-severe pain that affects daily life, inadequate relief from the above measures, and imaging that shows mild-to-moderate joint changes rather than complete loss of cartilage. A visible swollen knee — where fluid has accumulated — can be an additional signal; a 2026 multidisciplinary consensus noted that patients with synovitis or joint effusion may benefit from draining that fluid and following it with a steroid injection at the same visit.

Imaging findings alone do not qualify someone for an injection. Mild OA on an X-ray in the absence of significant symptoms is common, and the decision hinges on functional impact and treatment history, not the scan report.

Patients who are still in the early conservative phase — a few weeks into physiotherapy, or yet to try an appropriate oral anti-inflammatory — are not yet at this stage, and that is entirely normal.

Physical signs that point toward suitability

Certain physical features can help a patient gauge where they sit on the suitability spectrum before any clinic appointment — and some are visible without clinical training.

The clearest positive signal is fluid on the knee: a puffy, tight appearance around the kneecap that may worsen after activity or a night's rest. This is joint effusion, and it carries specific clinical weight. A 2026 multidisciplinary expert consensus recommended draining that fluid first and following it immediately with a corticosteroid injection — the combined approach addresses both the mechanical pressure of excess fluid and the inflammation producing it.

A knee that also feels distinctly warm to the touch, with an inflammatory quality to the flare rather than simple aching after exertion, points to active synovitis. This pattern is a good match for injection because the inflamed synovial lining is the direct target.

Mechanical wear signs are subtler but still informative. Morning stiffness that gradually loosens over less than an hour, a grinding or clicking sensation during movement (crepitus), and reduced ability to fully bend or straighten the knee are features a clinician will assess during physical examination. Together with pain that limits everyday activities despite adequate conservative care, they build the functional picture that supports moving to injection.

The opposite signal matters equally. When cartilage has worn away entirely — the bone-on-bone stage — gel-type injections such as hyaluronic acid are unlikely to provide meaningful benefit, as they rely on remaining joint space to function. Recognising these signs helps determine not just whether an injection is appropriate, but which type is most likely to fit.

Which injection — and why it depends on your OA stage

Not all injections work the same way, and the type that suits a patient is determined primarily by where their OA sits on the severity spectrum — as well as what they need from treatment.

Corticosteroid (steroid) injections are the first-choice option during an acute inflammatory flare: when the knee has become acutely painful, swollen, and limiting daily movement in a short-term spike rather than a gradual grind. They act quickly — typically within days to a week — and relief may last two to three months. Because repeated steroid injections can cause cumulative changes to joint cartilage, most clinicians limit them to approximately three to four per year.

Hyaluronic acid (HA) suits patients with early-to-moderate OA, particularly those who cannot tolerate or have not responded adequately to anti-inflammatory drugs. It functions as a lubricant within the joint and may improve mobility for several months. For patients in the UK, one practical distinction matters: NHS England guidance and NICE do not recommend HA for knee OA, so this option is accessed privately rather than through the NHS.

Arthrosamid, a polyacrylamide hydrogel, is aimed at patients with moderate-to-severe OA who are looking for longer-term non-surgical relief. Published data suggest symptom benefit may last up to three years, though long-term randomised trial data continue to accumulate. Suitability is assessed through a structured clinical framework at specialist level rather than as a routine first step.

PRP (platelet-rich plasma) is most commonly considered for active individuals with early-stage OA who are seeking a regenerative approach to managing inflammation — typically people who remain physically active, want to delay more invasive options, and are willing to engage with a treatment whose evidence base is still under active review. PRP is not FDA-approved for knee OA, and it is best understood as an option for patients who go in with clear expectations rather than a standard-of-care choice.

The OA grade on imaging — mild, moderate, or severe — is the primary anchor for this matching process. Two patients with knee OA may look very different clinically, and the injection suited to one may be inappropriate or ineffective for the other.

When an injection is not safe or unlikely to help

Safety stops and caution flags fall into two distinct categories, and knowing the difference helps patients arrive at their consultation better prepared.

Absolute contraindications — do not proceed

Certain situations make injection outright unsafe regardless of pain severity:

• An active skin infection, boil, or abscess on or near the knee
• Active systemic infection, septic arthritis, bacteraemia, or sepsis
• A prosthetic joint already in place in that knee
• An uncontrolled bleeding disorder

These are hard stops. Injecting through or near infected tissue risks driving bacteria into the joint — a potentially serious complication.

Relative cautions — clinical judgement required

Other situations call for careful assessment rather than automatic refusal:

• Poorly controlled diabetes: corticosteroid injections can temporarily raise blood sugar, sometimes significantly. This is not a permanent barrier, but glucose should be monitored for one to two weeks post-injection and the timing discussed with the treating clinician.
• Recent fracture within the knee joint: injection timing requires specialist review until healing is confirmed.
• Planned joint replacement in the near term: an injection shortly before surgery may complicate the procedure.

Additional factors that may raise infection risk — a history of prior knee surgery, rheumatoid arthritis, and advanced age — are each worth declaring at consultation.

One specific note on end-stage OA: complete cartilage loss affects suitability for gel-type injections in particular (hyaluronic acid and polyacrylamide hydrogel), not every injection type. Clinical trial data show poor outcomes in this group for viscosupplementation, and surgical options are usually the more appropriate pathway. Corticosteroid injection may still be considered for symptom management even at advanced stages, depending on the overall clinical picture.

Finally, some contraindications are not permanent. A resolved skin infection, for example, removes that specific barrier — but timing is always a clinical decision made at assessment.

What a suitability assessment actually involves

Arriving at a consultation expecting to walk out with an injection the same day is common — and usually inaccurate. The assessment that precedes any knee injection follows a consistent sequence, each stage serving a distinct purpose.

Clinical history comes first. The clinician will ask how long the pain has been present, how it behaves through a typical day, which activities it disrupts, and — crucially — which treatments have already been tried and how they worked. A patient who has not yet completed a genuine trial of physiotherapy or oral analgesia may be redirected before injection is considered.

Physical examination follows: range-of-motion testing, checking for crepitus (a grinding or creaking sensation in the joint), tenderness mapping, and assessment of whether fluid is present. Morning stiffness duration is also noted — in knee OA it typically resolves within 30 to 60 minutes.

Imaging, when needed, begins with X-ray, which shows joint-space narrowing, osteophyte formation, and subchondral changes that allow OA to be graded. MRI is used when the clinical picture remains unclear, adding resolution on cartilage quality, soft-tissue structures, and bone oedema. A scan that shows OA is not a decision in itself — it is one input alongside symptoms and examination findings.

The assessment runs in two steps: ruling out contraindications, then matching the appropriate injection to the patient's OA grade, symptom pattern, and treatment history. The same clinical data that confirm suitability also guide which option is recommended.

Realistic expectations: how long do injections last, and what do we still not know?

Duration of benefit is where injection science is most honestly incomplete — and patients deserve to know that upfront.

For corticosteroids, the ceiling on repeat use is the more practically important fact: cumulative effects on joint cartilage, not a limit on short-term efficacy, are what cap frequency at three to four injections per year. A patient planning long-term symptom management on steroid injections alone is working against the evidence.

For hyaluronic acid, the most active debate is not whether it works but how many doses produce the best result. Recent meta-analyses show no consistent outcome difference between a single injection and a course of three to five — the optimal regimen remains an open question, not a settled one.

For Arthrosamid, long-term randomised trial data are still being collected. Describing symptom benefit at around three years as an emerging signal is accurate; describing it as an established result at that timeframe is not yet supported by the published trial record.

For PRP, results vary considerably across studies, and no regulatory body has specifically approved it for knee OA. These are not peripheral caveats — they are the current state of the evidence.

Across all options, individual response varies more than suitability criteria can fully predict. Joint anatomy, inflammation pattern, and biological response all affect outcomes in ways that differ from patient to patient. Suitability increases the probability of benefit; it does not guarantee a particular result. When the evidence base is genuinely incomplete, the most useful next step is a structured clinical assessment that maps your specific profile against what the evidence does clearly show.