Osteoarthritis of the knee is among the most disabling musculoskeletal conditions worldwide, and its burden rises steadily as the population ages and as obesity becomes more prevalent. Estimates from the Global Burden of Disease have placed hip and knee osteoarthritis among the leading contributors to years lived with disability, and the knee accounts for the greater share of that burden.(1) The medial tibiofemoral compartment bears most of the load transmitted through the joint during walking, which explains why isolated medial disease is the most common radiographic pattern in clinical practice. The external knee adduction moment, the product of the frontal plane ground reaction force and its lever arm relative to the knee joint centre, has long served as the principal biomechanical surrogate of medial compartment loading, because a higher moment reflects a more medial position of the resultant force.(2) The relationship between this moment and the patient experience of pain is, however, modest and inconsistent, a point that has been quantified in recent syntheses of the biomechanical literature.(3)

Lateral wedge insoles and foot orthoses are inexpensive, non invasive devices that were designed to unload the medial compartment through the foot. A lateral wedge tilts the foot into eversion, moves the centre of pressure toward the lateral side of the shoe and shortens the frontal plane lever arm of the ground reaction force, all of which lower the external knee adduction moment measured in the gait laboratory.(4) Foot orthoses prescribed for medial knee osteoarthritis usually combine a lateral wedge with a medial arch support, a subtalar strap or, more recently, custom geometry produced by three dimensional printing, and each of these design choices is expected to modify both the magnitude and the inter individual reproducibility of the unloading effect.(5,6) The theoretical appeal of the approach is strong: because the wedge acts upstream of the knee through the foot, it does not require muscle strength or adherence to an exercise programme, and it can be worn throughout the day inside ordinary footwear. The trade off is that the same wedge that unloads the medial compartment also everts the ankle and rearfoot, which may displace load toward the lateral ankle structures or toward the patellofemoral joint, and this is one of the mechanisms proposed to explain why some patients do not tolerate or do not benefit from the device.(7)

The published evidence is discordant, and this is precisely why the intervention remains controversial. Meta analyses of randomised trials concluded that lateral wedge insoles do not produce a clinically meaningful reduction in knee pain, even though they consistently reduce the adduction moment in laboratory studies.(4,5) Earlier systematic reviews reached similarly cautious conclusions about the efficacy of laterally wedged insoles,(8) and a systematic review restricted to placebo controlled comparisons reached the same conclusion.(9) The Cochrane review of braces and orthoses judged the evidence to be of low certainty.(10) Large randomised trials found no symptomatic advantage of a lateral wedge over a neutral insole at twelve months,(11) and early crossover studies from the preceding decade had already suggested that the pain benefit was at best modest.(12,13) A more recent trial that first excluded biomechanical non responders before randomisation reported a small but statistically significant pain benefit,(14) which re opened the debate about whether the devices fail on average because they are offered to the wrong patients. Since 2015 the literature has therefore shifted from the question of whether lateral wedge insoles work on average toward two narrower questions, namely which patients respond and which insole design achieves the most reliable unloading, and this recent evidence has not been brought together in a synthesis confined to the past decade. The same period saw the spread of responder analysis, three dimensional printed orthoses and wearable sensors for gait monitoring, developments that make a dedicated synthesis of the 2015 to 2024 evidence timely even though several earlier reviews exist.

The objective of this systematic review was to synthesise the evidence published between 2015 and 2024 on the effects of lateral wedge insoles and foot orthoses in adults with medial knee osteoarthritis. The review had three specific objectives: to describe the design, population and methodological quality of the included studies; to determine whether these devices reduce self reported knee pain and physical function; and to determine whether they reduce the knee adduction moment and related surrogates of medial knee load.

This systematic review was conducted and reported according to the PRISMA 2020 statement.(15)

Studies were eligible when they enrolled adults with medial tibiofemoral knee osteoarthritis; evaluated a lateral wedge insole or a foot orthosis intended to unload the medial compartment; compared the device with a neutral or flat insole, a sham device, no device, usual care or another orthosis; and reported at least one of knee pain, self reported physical function or the knee adduction moment or a related measure of medial load.

Randomised trials, crossover trials, quasi experimental studies and prospective cohorts were included. Reviews, protocols, editorials, case reports, computational studies without human participants and studies that enrolled only healthy participants were excluded. Publications had to fall between 2015 and 2024 and to provide an English or Spanish title and abstract. Studies were not excluded on the basis of the wedge angle, the type of foot orthosis or the choice of comparator, so that the full range of designs used in clinical practice was captured.

Scopus, PubMed, OpenAlex and Crossref were searched. The core search combined population and intervention terms with a publication year filter, for example in Scopus: TITLE-ABS-KEY("lateral wedge" AND "knee osteoarthritis") AND PUBYEAR > 2014 AND PUBYEAR < 2025. Equivalent queries used "foot orthoses", "laterally wedged", "knee adduction moment" and "valgus insole" as variants, and the same strategy was adapted to the syntax of each database. In full, the five core queries were TITLE-ABS-KEY("lateral wedge" AND "knee osteoarthritis"), TITLE-ABS-KEY("laterally wedged" AND knee), TITLE-ABS-KEY("foot orthoses" AND "knee osteoarthritis"), TITLE-ABS-KEY("knee adduction moment" AND (insole OR orthoses OR wedge)) and TITLE-ABS-KEY("valgus insole" AND knee), each combined with the publication year filter. Supplementary searches for contextual and methodological literature used the terms "medial knee osteoarthritis", "knee adduction moment", "meta-analysis" and "biomechanics" without a lower date limit.

One reviewer extracted, for each included study, the design, country, sample size, population, intervention, comparator, outcomes and the main result with its estimate, using the full text when it was openly accessible and the abstract otherwise. Data were recorded in a structured table and were not re extracted independently, so extraction errors cannot be excluded.

The extraction form recorded first author and year, country, design, sample characteristics, the insole or orthosis specification including wedge angle, arch support, subtalar strap and manufacturing method, the comparator, the outcome measures and their time points, and the main quantitative result. Percentage changes in the adduction moment and in pain scores were transcribed as reported by the authors and were not recalculated. When a study reported several outcome domains, the pain, function and knee adduction moment results were prioritised for the narrative synthesis.

Randomised trials were assessed with the revised Cochrane risk of bias tool, RoB 2.(16) Crossover and quasi experimental biomechanical studies, cohorts and cross sectional analyses were assessed with a domain based checklist derived from the JBI appraisal instruments, covering selection, confounding, measurement and completeness of outcome reporting. Each study received an overall judgement of low, some concerns or high risk of bias. No study was excluded solely for risk of bias.

The expectation that a demonstrable reduction in the knee adduction moment would translate into a proportionate reduction in knee pain was only partly confirmed. Lateral wedge insoles and foot orthoses unload the medial compartment of the knee reliably and in a dose dependent manner in the laboratory, but the magnitude of unloading is small, it is not reproduced in every participant and it does not translate into a consistent symptomatic benefit. This contrast between a robust biomechanical signal and a weak clinical signal is the central finding of the review, and it is the same contrast that the field has been trying to reconcile since the first randomised trials of wedged insoles were published.(12,13)

The biomechanical results agree with earlier meta analyses, which estimated small immediate reductions in the first peak of the adduction moment of roughly 5 to 8 percent with lateral wedge insoles,(5,71) and with studies showing that arch supported and higher angle wedges augment the effect.(7,72) The absence of a consistent relationship between static foot posture and the biomechanical response, observed in the included studies, agrees with the broader literature on footwear and insoles, which has repeatedly failed to find simple anthropometric predictors of unloading.(51) The present review adds the observation, derived from studies published after 2015, that the dose response extends to custom and three dimensional printed orthoses,(41,43) and that a sizable minority of patients are biomechanical non responders in whom the moment increases rather than decreases.(45,46) This heterogeneity is not random. Responders tend to walk faster, to be female, to have less varus and less severe disease, and to have limited frontal plane ankle mobility, whereas participants with greater hindfoot flexibility unload the knee more with an ankle foot orthosis than with an insole.(48,49) The same individualisation was anticipated by early work that titrated the wedge angle to the patient(73) and by comparisons showing that braces and insoles do not unload the knee equally.(74) These modifiers explain part of the discrepancy between laboratory efficacy and the null results of clinical trials. A further source of inconsistency lies in the comparison condition and the duration of follow up. Studies that used a neutral insole sham found smaller effects than those that compared the wedge with no treatment, and immediate laboratory measurements, which are insensitive to habituation and to clinical relevance, overstate the benefit relative to measurements taken after weeks of wear. The outcome instrument matters as well. The adduction moment responds to a wedge within minutes, whereas pain and function integrate loading, inflammation, central sensitisation and expectation over time, so a concordance between the two is not guaranteed and was in fact rarely observed in the included studies.(3) The finding that the unloading effect can disappear after six weeks while symptoms remain stable illustrates how the early biomechanical response cannot be read as a proxy for clinical outcome.(47)

The pain findings align with the conclusion of the Cochrane review and of the largest meta analysis that a lateral wedge insole does not confer a clinically meaningful average pain benefit over a neutral insole,(4,10) and with a systematic review restricted to placebo controlled comparisons.(9) They also extend the negative result of the twelve month Bennell trial, in which a lateral wedge was no better than a control insole, to the smaller trials published over the past decade.(11,17,18) The one prescreened trial that reported a statistically significant advantage found a between group difference of only 0.7 points on a 0 to 10 scale, an effect that its authors considered clinically relevant in only a minority of patients.(14) Across the placebo controlled comparisons, the point estimates for pain clustered around zero to one point on the numerical scales, and none of the differences reached the commonly proposed minimal clinically important difference for knee osteoarthritis. Studies that lacked a neutral insole control reported larger and more optimistic improvements,(20,26) which illustrates how the absence of a sham comparator inflates the apparent benefit of a device whose main effect may be contextual. Whereas the neutral insole arms of the randomised trials also improved, the uncontrolled studies could not separate the wedge from attention, expectation and the passage of time. The narrative reviews and year in review summaries of the period reached broadly the same reading of this literature.(6,75,76) Taken together, the certainty of the evidence for an average pain benefit is low, because the randomised comparisons are few, mostly unblinded and imprecise, whereas the certainty of the biomechanical evidence is moderate, because the direction of the adduction moment reduction is reproducible across independent laboratories even though its magnitude is modest. This asymmetry in certainty between the biomechanical and the clinical questions is itself informative: it indicates that the failure of the device is not primarily a failure to unload the knee, but a failure of unloading, at the magnitude that is achieved, to translate into symptom relief. A similar reading emerges from the studies that tested the insole and a valgus knee brace head to head, in which the brace generally unloaded the knee and relieved pain more than the insole,(32,74,77) which suggests that the marginal clinical value of the insole is real but small relative to other unloader devices.

For future trials, three design features would materially improve the evidence. First, the comparator should always be a neutral insole of identical appearance, because uncontrolled and brace comparison designs overstate the benefit. Second, the outcome set should pair a patient reported pain measure with the adduction moment measured before and after a habituation period, so that the biomechanical and clinical responses can be examined together in the same participants. Third, the analysis should be stratified by the baseline variables that predict response, particularly varus alignment, hindfoot mobility and gait speed, rather than reporting only the group mean. Trials designed in this way would answer the question that the current literature only begins to address, which is not whether the average patient benefits but which patient benefits.

For clinical practice, the findings support a cautious and selective use of lateral wedge insoles. The devices are inexpensive and safe, and they can be offered to patients who understand that the expected pain relief is small, that benefit is more likely in early disease and in less varus knees, and that a proportion of patients will not unload the knee at all. The prescription should favour a full length wedge of at least 5 degrees, with an arch support when comfort requires it, and should avoid arch support alone, which can increase loading in some patients.(54) Unilateral use should be preferred when the contralateral knee is less symptomatic, given the adverse impact loading observed with bilateral eleven degree wedges.(56) The dose response to wedge angle also suggests that the angle should be titrated to the individual rather than applied as a fixed prescription, a principle that was proposed in early studies of subject specific wedging(73) and that the more recent predictive work on responders now makes actionable.(48) For research, the priority is to replicate the prescreening approach, in which a gait based confirmation of unloading precedes randomisation, and to test whether predicted responders achieve clinically meaningful pain relief in adequately powered trials.(14,45) The same logic applies to the newer custom and three dimensional printed designs, which promise personalisation but have not yet been shown to outperform standard wedges on symptoms.(41,43) Because the expected benefit is modest and the devices are low cost and low risk, shared decision making is appropriate, with the patient informed that the wedge is a trial of therapy whose value should be judged against individual symptoms after a few weeks rather than assumed in advance.

The available evidence indicates that lateral wedge insoles and foot orthoses reduce the knee adduction moment and medial knee load immediately and in a dose dependent manner in adults with medial knee osteoarthritis, although the effect is small, variable and not maintained in every participant. These devices provide only a small and inconsistent reduction in knee pain and no consistent advantage in physical function over neutral insoles, and the symptomatic benefit appears confined to identifiable subgroups. Unloading of the medial meniscus and improvements in postural control are reported mainly in early disease, which supports the selective, personalised use of these orthoses rather than their routine prescription.