Why physiotherapy comes first — and why six weeks

Most people referred to physiotherapy for knee osteoarthritis arrive with a quiet suspicion: that this is a formality — something to complete before a scan or injection becomes available. That suspicion is understandable, but it misreads what the evidence actually says.

NICE guideline NG226 (2022) does not recommend structured exercise as a first step because it is cautious about other options. It recommends it because exercise is the only intervention with Level 1a evidence of benefit at this stage of the disease. OARSI and EULAR carry the same designation — this is not a UK-specific policy position but a global clinical consensus built on meta-analyses of randomised controlled trials. The guideline explicitly places imaging, early analgesia, glucosamine, and injectable therapies further along the pathway, not alongside exercise.

Six weeks is the chosen horizon for a specific clinical reason: it is long enough to produce measurable change in pain, strength, and function, and therefore long enough to give a specialist meaningful information about how a patient is responding before considering whether to escalate. In practice, a structured programme typically involves supervised sessions twice weekly — combining targeted quadriceps and lower-limb strengthening with functional movement — alongside a home exercise component. The six-week frame is a clinical decision point, not a waiting room.

What measurably changes in six weeks

The outcomes from a well-conducted six-week programme are specific enough to measure and track. Pain scores typically fall — not by trivial margins, but by amounts that reflect genuine change in daily function. Quadriceps and lower-limb strength improve in ways that are quantifiable on a leg-press one-repetition maximum test. Functional measures follow: the Oxford Knee Score, timed walk tests, and timed stair-climb assessments all show consistent improvement in the published literature.

That literature is substantial. A cumulative meta-analysis of 42 randomised controlled trials involving 6,863 patients, published in 2019, confirmed that exercise reduces knee OA pain post-treatment — with no evidence of publication bias and consistent effects in the studies at lowest risk of bias. The reviewers noted that the evidence of benefit had, in fact, been established as early as 2010; the decade since has reinforced rather than revised it.

For patients, these measures translate into concrete daily gains: getting up from a chair more readily, descending stairs without bracing, covering ground with less discomfort. For the clinician reviewing progress at the six-week mark, the same data — Oxford Knee Score, walk time, strength figures — become the basis for deciding whether the conservative pathway has delivered sufficient benefit or whether further assessment, injection therapy, or specialist review is warranted. Progress is not merely felt; it is recorded, compared against baseline, and used to shape the next clinical decision.

Why muscle strength is the biological priority

Patients often ask why a physio spends so much time on quadriceps and gluteal exercises — and comparatively little on stretching or heat. The answer lies partly in what OA pain does to the muscles surrounding the joint.

A 2025 CT study of 89 patients awaiting total knee replacement found that the more painful knee limb showed selective volume loss in the vastus lateralis, intermedius, and medialis — the muscles that form the front of the thigh — as well as the gluteus maximus. Most of these muscles also carried higher levels of intramuscular fat, a marker of disuse degeneration. The finding matters because it points in a specific direction: OA pain causes people to offload the joint, and that offloading progressively hollows out the very muscles designed to protect it. Targeted strengthening is not generic conditioning; it is correcting a documented deficit.

There is a second, separate rationale at cartilage level. Articular cartilage has no blood supply; it depends on synovial fluid — compressed in and pumped out by mechanical loading — for its nutrients and waste clearance. Think of the cartilage surface as a dense sponge: it only circulates fluid when it is squeezed. Cyclical weight-bearing exercise provides that compression, making controlled loading the biologically coherent way to support the cartilage environment, even when structural repair is not the primary goal.

These two mechanisms — reversing disuse atrophy, maintaining cartilage nutrition — provide a clear physiological basis for the exercise prescription. What they do not fully explain is how much of the clinical pain reduction exercise produces, and through what pathways. That picture is considerably more complicated, as the evidence examined in the next section shows.

The honest limits of what we know about how it works

Ninety-eight per cent of the reason exercise works for knee OA pain remains unexplained. That figure comes from the STEER OA study, a 2023 IPD mediation analysis that pooled 12 RCTs and 1,407 patients specifically to identify what is driving the clinical benefit. Quadriceps extension strength — the most intuitive candidate given everything established in the previous section — mediated only 2.3% of the effect on pain and 2.0% on function. Range of motion and proprioception were non-significant mediators. The effect is real; the mechanism is largely a black box.

More recent evidence recalibrates the expected scale of that effect. A 2026 systematic review in RMD Open (Schleimer et al.) compared exercise against placebo rather than no treatment — a methodologically tighter standard than most earlier trials used. Against that comparison, exercise provided only negligible-to-small benefits for pain and physical function. No-intervention controls conflate the effect of exercise with the effects of attention, expectation, and regular clinical contact. The Schleimer analysis separates them, and the resulting effect sizes are smaller.

A 2025 narrative review of five years of knee OA exercise research reached two further conclusions: no dose-response relationship between exercise and clinical outcomes has been demonstrated, and improving patient adherence to exercise does not reliably produce better results.

None of this overturns NICE NG226. The guideline mandate stands, the cumulative RCT base remains robust, and the recommendation is still unambiguous. The honest clinical position is that the benefit is real but more modest than older trials suggested, and the mechanism is unknown — which is a reason for clear expectation-setting at the start of a programme, not a reason to bypass it.

Who benefits most — and why early referral matters

Timing of referral turns out to matter more than most patients expect — and more than many GPs are told.

The OA Trial Bank's 2024 IPD meta-analysis pooled 10 RCTs and 1,767 participants to examine whether symptom duration at the point of referral predicts how much someone gains from exercise therapy. The result was striking: patients who had been symptomatic for one year or less benefited significantly more than those presenting later, both in the short term and at longer follow-up. The long-term pain difference — a mean difference of −8.33 on a 100-point scale in favour of earlier presenters — is a population-level estimate across trial participants, not a guarantee for any individual, but the interaction was consistent and statistically robust. The authors characterised it explicitly as a 'window of opportunity' for targeting exercise therapy early in the disease course.

For patients reading this, the implication reverses a common intuition. Many people with knee pain spend months or years self-managing — resting when it flares, modifying activity, hoping it settles — before seeking help. That approach is understandable, but this evidence suggests the physio programme begun in year one is likely to be more effective than the same programme begun in year four.

For referring clinicians, the OA Trial Bank finding provides a concrete basis for earlier referral. Structural severity — OA grade on imaging — was not identified as a consistent predictor of response in this analysis; symptom duration was the stronger signal. A patient with mild-to-moderate radiographic change who presents within a year of symptoms is, on current evidence, the patient most likely to respond.

Waiting to 'see how things go' is a reasonable instinct but a poorly timed strategy. Assessment and, where appropriate, early physiotherapy referral are not only guideline-concordant; they may represent the highest-yield point in the entire treatment pathway.

When six weeks is not enough — and what comes next

Completing six weeks of physiotherapy is a clinical milestone, not a verdict. For most patients it produces meaningful improvement — but not for everyone, and recognised non-response at the end of a well-structured programme is a clinical signal that warrants a different next step, not a reason for alarm.

One complication worth naming: fewer than 40% of patients with knee OA globally receive the first-line care that NICE and international guidelines specify. Passive treatments — heat, electrotherapy, and kinesiology taping — remain common in practice despite not featuring among NICE NG226's recommendations. Patients whose programme has been dominated by these approaches have not, in any clinical sense, completed a guideline-concordant course; active, progressive loading is the intervention the evidence supports.

Where a genuinely active programme concludes without adequate improvement in pain or function, specialist assessment is the logical next step. Its purpose is specific: to determine whether the joint remains a candidate for preservation-focused strategies, or whether structural progression means surgical planning — partial or total replacement — is the more appropriate frame. These are different pathways with different time-horizons and different practical implications, and imaging at this stage helps characterise which applies. Within the preservation-focused pathway, injection options such as corticosteroid, hyaluronic acid, or PRP may have a role alongside or instead of further exercise-based management; these are named here as context rather than recommendations, since individual suitability depends on clinical assessment.

For patients approaching this point, a joint-preservation consultation with a specialist — such as Professor Paul Y. F. Lee at the London Cartilage Clinic — can clarify whether the conservative window remains open, or whether the clinical picture calls for something more.

1. [1] Does osteoarthritis physiotherapy research in South Korea align with NICE guidelines: a systematic review. (2025). https://doi.org/10.1186/s41927-025-00496-w https://doi.org/10.1186/s41927-025-00496-w
2. [2] Recent highlights and uncertainties in exercise management of knee osteoarthritis. (2025). https://doi.org/10.1016/j.jphys.2025.06.010 https://doi.org/10.1016/j.jphys.2025.06.010
3. [3] Do we need another trial on exercise in patients with knee OA? No new trials on exercise in knee OA.. (2019). https://doi.org/10.1016/j.joca.2019.04.020 https://doi.org/10.1016/j.joca.2019.04.020
4. [4] Mechanisms of action of therapeutic exercise for knee and hip OA remain a black box: IPD mediation study with the OA Trial Bank. (2023). https://doi.org/10.1136/rmdopen-2023-003220 https://doi.org/10.1136/rmdopen-2023-003220
5. [5] People with short symptom duration of knee OA benefit more from exercise therapy than people with longer symptom duration: IPD meta-analysis from the OA Trial Bank. (2024). https://doi.org/10.1016/j.joca.2024.07.007 https://doi.org/10.1016/j.joca.2024.07.007
6. [6] Why muscle strengthening exercises should target the quadriceps and gluteus maximus in patients with knee OA. (2025). https://doi.org/10.1016/j.jot.2025.06.013 https://doi.org/10.1016/j.jot.2025.06.013