Stabilisation training for an unstable knee

A knee that gives way is one of the more frightening things a joint can do, and it is common in osteoarthritis. The instinctive prescription is to train the instability directly: balance work, proprioceptive drills, agility, perturbation — exercises that ask the knee to react to the unexpected and, in principle, teach it to hold.

The reasoning behind that is sound. Instability drives cartilage damage rather than merely accompanying it, and in animal models it is the mechanical consequence of a loose joint that starts the degeneration. What has not held up is the next step: that training aimed at instability fixes instability.

What instability means in these trials

Almost always, what the patient says. Instability in this literature is self-reported buckling or giving way, captured by interview or questionnaire.

That matters more than it first appears. Sensations of giving way, fear of giving way, and measurable joint laxity are three different things that a single interview question collapses into one. The systematic review's first recommendation to the field is to establish a proper method for assessing instability, dynamically and statically, noting that three-dimensional motion analysis can quantify it and that longitudinal studies of that kind are needed. A null result measured on an interview is weaker than a null measured on an instrument.

The six-study review

Searches across three databases returned fourteen candidate articles, of which six met the inclusion criteria: trials of exercise therapy aimed at joint instability in knee osteoarthritis, against general exercise therapy or usual care.

Exercise therapy worked, in the ordinary way — muscle strength and pain both moved. The instability-targeted component added nothing. Training focused on knee stabilisation, whether through proprioceptive, neuromuscular, agility or perturbation work, produced no additional benefit on instability ratings over muscle maintenance and strength training alone. The effects of specific instability training were not different from those of general exercise therapy.

Only two of the six studies could be pooled, and even those had a gap between their intervention periods.

The patients who did respond

One signal survives the null, and it is specific enough to act on.

Stabilisation therapy tended to help patients who reported many episodes of instability and already had strong muscles. The trial group behind that finding concluded that initial knee stabilisation therapy may have added value in exactly that combination. The review's inference runs the other way and is the practical one: for patients whose muscles are weak, strengthening first is what adds the value.

Read that as a sequencing result rather than a subgroup curiosity. Stabilisation work may need a base of strength underneath it to have anything to stabilise with, which would explain why trials that gave it to everyone found nothing on average.

Proprioception as a risk factor

A second line of evidence, from a different design, points the same way.

In a prospective cohort of community-dwelling adults aged 50 to 79 — 1,390 participants and 2,276 knees for the radiographic analysis, 1,829 and 3,166 for the symptomatic one — baseline knee joint position sense predicted neither incident symptomatic nor incident radiographic osteoarthritis over thirty months. The strength-by-proprioception interaction the study was built to test was null in both sexes and for both outcomes.

Knee-extensor strength, measured in the same people at the same visit, did predict incident symptomatic osteoarthritis: high versus low tertile, odds ratio 0.43, 95% CI 0.31 to 0.61.

The proprioceptive testing has real limits — seated, non-weight-bearing, right knee only — and seated joint position sense is not the same faculty as staying upright on a loose knee. But a trial-level null and a cohort-level null, generated by different methods, are harder to dismiss together than either alone.

What the strength evidence prescribes instead

The alternative is not "do nothing", and the evidence for it is much stronger than the evidence against stabilisation work.

Land-based therapeutic exercise reduced knee osteoarthritis pain immediately after treatment across 44 trials and 3,537 participants — a standardised mean difference of −0.49, about 12 points against a control-group pain of 44 on a 0–100 scale, rated high-quality evidence. Physical function improved across 44 trials and 3,913 participants. Benefit persisted at least two to six months after the formal programme ended, and people dropped out of the exercise arms no more often than the controls.

The reviewers make no recommendation about optimal frequency, intensity or duration, because the programmes were too heterogeneous to support one. What the literature supports is that the leg gets stronger and the knee hurts less, which is the base the instability finding says to build first.

Where the review could not look

Three variables that could plausibly carry the instability signal were outside the review's scope: fear of movement, varus-valgus alignment — the lateral thrust of a knee that shifts sideways under load — and whether one knee or both were affected.

Alignment in particular is a mechanical cause of a knee giving way that no amount of proprioceptive drill would address, and it belongs to a different treatment pathway. Publication bias could not be assessed at all, because too little has been published on instability in knee osteoarthritis to test for it.

Open questions

The trial to run is the one the review implies: instability measured on an instrument rather than an interview, with participants stratified by baseline strength, comparing strengthening followed by stabilisation against either alone. That design would test the sequencing hypothesis directly, and it is the hypothesis the surviving subgroup points at.

Until then, the actionable reading is a positive one. The intervention with the largest evidence base in knee osteoarthritis — getting the leg strong — is also the one the instability literature keeps pointing back toward, and for a knee that gives way it is the place to start rather than a consolation prize.