How Hard Can You Push Exercise in Knee OA? The Evidence Is Less Settled Than the Guidelines Imply

Let's be upfront: in the dose-response meta-analysis below, the certainty of the evidence on resistance exercise dosing in knee osteoarthritis was low to very low. The studies below range from a meta-analysis with low- to very low-certainty evidence to a small 8-week trial and observational cohorts that show associations only. What follows is not a new prescription algorithm — it's a picture of what the literature currently does and doesn't say about intensity, volume and impact, so you can weigh it against what the guidelines imply.

Lighter loads, more repetitions — with caveats

Núñez-Cortés et al. (2026) systematically reviewed 20 RCTs (n = 791) in people with knee osteoarthritis, comparing different intensities and delivery methods of isotonic resistance exercise. Very light- to moderate-intensity isotonic exercise produced the greatest reductions in pain, while moderate intensity produced the greatest improvements in function.

Before you rebuild your programmes around those numbers, note the authors' own caveat: the certainty of evidence was low to very low. Treat the estimated peaks — pain reductions at 6,000 repetitions and functional improvements at 2,600 repetitions of moderate intensity — as hypotheses to observe in your own patients, not as targets.

Moderate resistance vs. blood flow restriction

Tahoon et al. (2026) randomised 75 patients with knee osteoarthritis, aged 45–60, to passive BFR at 50% LOP, passive BFR at 70% LOP, or moderate-intensity resistance training (MI-RT), with outcomes assessed at week 8. At week 8, MI-RT produced greater increases in quadriceps torque than both BFR groups. But BFR-70% LOP showed greater improvement in lower-extremity functional ability than MI-RT. And no significant between-group differences were observed for pain intensity, KOA impact, functional mobility, or perceived improvement.

In other words, the modality that won on torque did not win on function or pain — a useful reminder that your chosen outcome measure determines which intervention looks better.

High-impact activity and surgery: an association, not a licence

In a longitudinal mediation analysis of GLA:D registry data, Liew et al. (2026) found that baseline high-impact physical activity was associated with 36–48% lower odds of total hip replacement at 12 months. In participants with knee OA, high-impact physical activity was not associated with altered risk of total knee replacement. A small indirect effect via functional capacity was also observed, where greater functional capacity was associated with a reduction in odds of TKR (OR 0.92, 95% CI 0.87 to 0.98).

This is an observational association in a mixed hip and knee population — it does not establish that pushing patients into high-impact activity protects their joints. It does, however, give you something concrete to say to the patient who believes impact will inevitably wear their joint out.

Where this leaves your programming

For context: a narrative review of hip and knee OA found no exercise type proven superior and no established dose-response relationship, with exercise therapy providing small to moderate improvements in pain and function. Neuromuscular training showed significant benefits over control for KOOS Symptoms, ADLs and QoL, but not for Pain or Sport/Rec. And a retrospective cohort study of patients with knee OA, with and without type 2 diabetes (1,329 matched pairs with T2DM and 3,441 matched pairs without), found exercise therapy associated with lower TKA utilisation in the KOA-only cohort but not in the T2DM+KOA cohort, with increased injury risk in the latter — though the authors state causal treatment-effect modification cannot be inferred.

A reasonable reading for practice: moderate-intensity isotonic work is a defensible default for pain and function; passive BFR at 70% LOP remains an option when loading is not tolerated, with functional gains possible even when torque lags; and the fear of impact in this population is not well supported by this 12-month registry cohort. Beyond that, the honest answer is that the dose question is still open — no exercise type has proven superior and no dose-response relationship has been established — so your own outcome measures at follow-up remain a reasonable guide for adjusting the programme.

Since the dose question remains open, PhysioSync's Home Programs let you prescribe and schedule the programme you choose, while PainSync tracks each patient's pain and functional scores over time, so progression follows what each patient actually shows.