Digital support, water, eccentric loading: what knee OA exercise evidence actually shows

If you have been trying to work out whether home-based, digitally supported or supervised exercise is the better prescription in knee osteoarthritis, the honest answer is that the incremental value of synchronous or hybrid professional support over face-to-face supervised exercise remains uncertain. What follows comes from three 2026 reviews and one trial, and — as you will see — the certainty behind many of the between-group differences is low or very low, and several effects are small or imprecise. This is a short read, because there is not much more to say yet.

Digital support: function differences, pain less clear

Luo et al. (2026) ran a network meta-analysis of 2,250 adults with knee osteoarthritis, comparing seven nodes of exercise and self-management strategies classified by delivery, feedback and professional involvement. For physical function, two of the three digitally supported comparisons (D1 vs. C1 and D2 vs. C2) showed improvements over their comparators (D1 vs. C1: g = 0.47, 95% CI 0.31 to 0.64; D2 vs. C2: g = 0.47, 95% CI 0.22 to 0.72), with moderate-certainty evidence, while the D3 vs. C3 estimate was imprecise (g = 0.48, 95% CI -0.13 to 1.10; low-certainty evidence). The authors stress that these contrasts used different comparators and do not establish superiority of digital delivery itself. All three pain comparisons also favoured the digitally supported nodes (g = 0.15, 0.21 and 0.50), but every 95% CI included the null value, with low-certainty evidence.

The authors are careful about what this does not mean: the incremental value of synchronous or hybrid professional support over face-to-face supervised exercise remains uncertain, and current evidence does not support prioritising it solely on the basis of rankings or assuming that more digital support means more effect. Retention differences for D2 vs. C2 and D3 vs. C3 were likely small, and local inconsistency was identified for D1 vs. C1; retention estimates were generally close to the null.

In a related population of patients with osteoarthritis and inflammatory arthritis, Zhang et al. (2026) reviewed randomised trials of telemedicine interventions (24 RCTs, n = 2,262, of whom 1,568 had OA) and found associations with improvements in pain, quality of life, WOMAC and self-efficacy — but with I² above 75% on all primary outcomes and 95% prediction intervals spanning the null, the authors call these findings exploratory.

Aquatic versus land: small or imprecise

Amaral et al. (2026) systematically reviewed supervised aquatic exercise against supervised land-based exercise in 700 adults with knee osteoarthritis. Differences in VAS pain generally ranged from -1.80 to -0.32 scale units in favour of aquatic exercise. Evidence for physical function and quality of life was inconsistent, and the observed differences were generally small or imprecise. The authors note that clinical importance remained uncertain because applicable thresholds were not available for all measures, and that certainty of evidence ranged from very low to high, predominantly low or very low. If you have a pool and a patient who prefers it, this does not argue against aquatic work — but it does not establish superiority either.

Adding eccentric work: the clearest signal

Hossain et al. (2026) randomised 120 people with knee osteoarthritis to eccentric exercise plus a multimodal intervention (n = 60) or the multimodal intervention alone (n = 60), both supervised for 6 weeks, with follow-up at week 30. At week 6, between-group differences favoured the eccentric-plus-multimodal group for NPRS (MD -0.570, 95% CI -1.003 to -0.136), PPT (0.225 kg/cm², 0.131 to 0.319), WOMAC pain (-1.145, -1.929 to -0.361), function (-2.646, -4.708 to -0.584), 6MWT (19.347 m, 14.743 to 23.951) and PHQ-9 (-1.136, -1.877 to -0.394). At week 30, these benefits were sustained or increased, with additional improvement in knee-flexion ROM (3.139°, 1.128 to 5.150), although knee-extension ROM did not differ.

A reasonable starting point from all this: keep prescribing exercise — these reviews do not establish that one delivery format is clearly better than another — and where eccentric loading is feasible and tolerated, this trial gives the strongest between-group signal of the set. One RCT of 120 participants is not a practice-changing foundation, but it is the finding here most worth watching.

Whichever format you choose, PhysioSync's Home Programs put the prescribed exercises in the patient's calendar and track adherence between visits, while you decide which mix of supervised and home work suits each person.