The effect of a non-invasive biomechanical treatment on subjective and objective measures in patients with knee osteoarthritis- A retrospective analysis on a UK population


 Background Osteoarthritis is a major cause of pain and disability worldwide, therefore ways of treating this condition are paramount to a successful health system. The purpose of the study was to investigate the changes in spatial-temporal gait parameters and clinical measurements following treatment with a non-invasive biomechanical device on patients with knee osteoarthritis within the UK. Methods A retrospective analysis was carried out on 455 patients with knee osteoarthritis. All patients were evaluated using a computerized gait test and two self-assessment questionnaires (WOMAC and SF-36) at baseline and after 3 and 6 months of treatment. The biomechanical device was individually calibrated for each patient to minimise symptoms whilst walking and train neuromuscular control. Patients used the device for short periods during activities of daily living. Repeated measures statistical analyses were performed to compare differences over time. Results After 6 months of therapy significant improvements were seen in all gait parameters (p<0.01). Specifically, gait velocity, step length and single limb support of the more symptomatic knee improved by 13%, 7.8% and 3%, respectively. These were supported by significant improvements in pain, function and quality of life (48.6%, 45.7% and 22% respectively; p<0.001). A sub-group analysis revealed no baseline differences between those who were recommended joint replacement and those who were not. Both groups improved significantly over time (p<0.05 for all). Conclusions Our results suggest the personalised biomechanical intervention can improve gait patterns, pain, function and quality of life. It may provide an additional solution to managing UK patients suffering from knee osteoarthritis.

Background Osteoarthritis is a major cause of pain and disability worldwide, therefore ways of treating this condition are paramount to a successful health system. The purpose of the study was to investigate the changes in spatial-temporal gait parameters and clinical measurements following treatment with a non-invasive biomechanical device on patients with knee osteoarthritis within the UK. Methods A retrospective analysis was carried out on 455 patients with knee osteoarthritis. All patients were evaluated using a computerized gait test and two self-assessment questionnaires (WOMAC and SF-36) at baseline and after 3 and 6 months of treatment. The biomechanical device was individually calibrated for each patient to minimise symptoms whilst walking and train neuromuscular control. Patients used the device for short periods during activities of daily living. Repeated measures statistical analyses were performed to compare differences over time. Results After 6 months of therapy significant improvements were seen in all gait parameters (p<0.01). Specifically, gait velocity, step length and single limb support of the more symptomatic knee improved by 13%, 7.8% and 3%, respectively. These were supported by significant improvements in pain, function and quality of life (48.6%, 45.7% and 22% respectively; p<0.001). A sub-group analysis revealed no baseline differences between those who were recommended joint replacement and those who were not. Both groups improved significantly over time (p<0.05 for all). Conclusions Our results suggest the personalised biomechanical intervention can improve gait patterns, pain, function and quality of life. It may provide an additional solution to managing UK patients suffering from knee osteoarthritis.
Background Osteoarthritis (OA) is a major cause of pain and locomotor disability worldwide. The knee is the most commonly affected weight-bearing joint with 4.11 million people having knee OA in England (1). With an aging population and a rise in obesity being leading risk factors (2), it is estimated that by 2020, the number of people suffering from knee OA will rise to 6.5 million (1). The condition leads to social, psychological and economical burdens with substantial financial consequences (3). Therefore, cost-effective and non-invasive ways of treating and managing this condition more effectively are paramount to a successful health system.
Currently, the National Institute of Clinical Excellence (NICE) guidelines outline core treatments such as education and exercise as first line care, progressing to more advanced biomechanical modalities such as valgus knee braces and orthotics in some cases (4). Total knee replacement (TKR) is the most common treatment for end-stage knee OA and appears to be increasing over-time (5) with studies estimating that the rates of TKA will reach 119,000 procedures per year by 2035 (6). TKR's have revolutionised the care of patients with knee OA and are considered an effective intervention for the treatment of chronic knee pain and disability (7). However, as well as the associated issues of surgical intervention, there is also evidence based on using patient reported outcome measures (PROM's), that some patients experience chronic knee pain, functional disability, and poor quality of life after TKR (8). It is suggested approximately 18% of patients report the outcomes of their surgery as only fair or poor, with a small proportion of these experiencing complications (9).
One factor that has been heavily researched over the past few decades is the effect of lower limb biomechanics upon the development and progression of the disease (10,11).
Patients with knee OA often present changes in spatial-temporal gait parameters including a reduction in gait velocity, step length and single limb support phases. Moreover, these changes were found to correlate with the levels of pain and functional disability (12,13).
External knee adductor moment (KAM) and knee adductor angular impulse (KAAI) have been suggested as the surrogate kinetic variables for expressing medial joint forces and overall cumulative loading of the knee throughout the stance phase respectively (11,14,15). Neuromuscular changes related to knee osteoarthritis such as muscle weakness and altered muscle activation have also been widely acknowledged as contributing factors to the development of OA (11,(16)(17)(18)(19).
Symptomatic knee OA has been shown to be influenced by both biomechanical and neuromuscular treatments, which has led to conservative treatments currently available largely aim to influence either one of these factors (20)(21)(22)(23)(24)(25)(26)(27). Interventions typically attempt to manipulate the KAM and KAAI loads transmitted through the joint, or improve the neuromuscular deficits seen within the pathology by muscle strengthening and proprioceptive exercises (28,29). However, these modalities often occur in isolation and can reduce effect in time, which may bring into question their effectiveness, leading to further deterioration of symptoms and eventually surgery (22,30,31).
For the past decade, a personalised non-invasive biomechanical treatment for patients with knee OA has been available in the UK. The extensive evidence about the effect of this treatment on clinical outcomes of patients with knee OA is promising (31-35), albeit no research is yet published on UK populations. Therefore, this study aims to review the effect of this therapy on UK populations suffering with knee OA.

Methods
A retrospective chart review analysis was carried out to examine the effect of the biomechanical therapy upon spatial-temporal gait parameters, levels of pain and function, along with perceptions of quality of life in patients suffering from knee OA. The protocol was approved by the University of Roehampton Ethics Committee and all participants provided consent for their data to be used within in the study. A search for eligible data was conducted on the database of treatment centres providing the biomechanical therapy between 2009 and 2017. Figure 1 represents the data reduction flow chart and the eligibility for the study against the inclusion/exclusion criteria.
Four hundred and fifty-five patients, 247 females (54%) and 208 males (46%) with symptomatic knee OA participated in this study. Mean (SD) age was 61.6 (8.7) years and 62.9 (10.3) years, respectively. All patients completed both computerised gait analysis and validated PROM's at initial consultation and at 3 and 6 months after commencing the therapy. The OptoGait system (Version 1.11) was used to measure spatial-temporal gait parameters at each data collection point (36). It works by detecting the interruption between the transmitting and a receiving LED (light emitting diode) bars. This interruption produces quantifiable data that can highlight differences between normal to pathological gait such as in knee OA (13,37). Patients walked at a self-selected speed over a 4 m measurement area, with 2 metres before and after to allow for constant velocity walking speed during data capture. Each gait test included 4 lengths in which the mean values were calculated for each parameter. The parameters recorded include velocity (cm/s), step length (cm) and single-limb support (SLS) phase (% gait cycle).
Step length and SLS was calculated for the less and more symptomatic limbs respectively. The more symptomatic limb was determined by the lower SLS at baseline. All gait assessments were conducted in barefoot at each data collection point.
The Knee Osteoarthritis Function Grade (KOFG) was used to classify patients' and assess improvements in gait over the treatment period. It is a validated classification tool utilising cadence and stride length from the spatial-temporal gait analysis as predictor variables to assess changes in functional severity (13). The KOFG is a four-grade scale with 1 being the best function and 4 the worst function. According to the model, a shorter stride length with lower cadence is indicative of a higher functional severity grade disease (more severe knee OA), while a longer stride length with higher cadence is indicative of a lower functional severity grade of disease (less severe knee OA).
The WOMAC questionnaire was used to evaluate changes in patients' perception of pain and function (38). It contains 24 visual analogue scale (VAS) questions that can be divided during gait with the goal to re-orientate the GRF vector and to reduce the loads in the affected area of the joint while walking (40)(41)(42). The convex nature of the elements induces a level of controlled perturbation and proprioceptive training causing muscles in the lower limb to work differently (43,44).
All patients received instructions to wear the biomechanical device during walking in his/her daily routine (at home/work), for half an hour a day at the start of the treatment, gradually increasing (+ 10 min/week) up to 3 hours/day after 3 months. Patients undertook follow-up assessment and re-calibration of the device approximately 3, 12, and 24 weeks after commencing the treatment in-order to optimise the pain alleviating characteristics of the device and progress the convexity to induce further neuromuscular challenges as required.
Data was analysed with SPSS software version 23.0 with significance levels set at p < 0.05. Data is presented as mean and standard deviations for gait spatial-temporal parameters and questionnaires (WOMAC/SF-36), followed by a 95% confidence interval (CI) for all time periods. The Kolmogorov-Smirnov non-parametric test was used to establish the normal distribution of spatial-temporal parameters and therefore use parametric statistical tests. To assess for the potentially confounding effects of age and gender in the WOMAC and SF-36 scores, analysis of covariance was conducted.
WOMAC and SF-36 were compared using the non-parametric equivalent methods of Kruskal-Wallis test to compare all three groups, and Mann-Whitney U-tests for pairwise comparisons. Repeated measure ANOVA's were performed for the differences between the outcomes, measured at baseline, 3 and 6 months into the therapy. Multiple linear regression analysis was used to address the percentages changes after 6 months. The WOMAC and SF-36 scores were the dependent variables, with the gait parameters used as independent variables. Because age is typically related to PCS and gender to MCS, the linear regression models initially included age and gender. Interaction terms were also included in the regression model. To adjust for the potentially confounding effects of socio-demographic factors (age and sex) in the SF-36 and WOMAC scores, analysis of covariance was conducted. All reported p-values are from two-sided tests.

Results
All spatial-temporal gait parameters significantly improved following 3 months of therapy (all less than p < 0.01). There were also further significant improvements in all parameters between 3 and 6 months of therapy (All less than p < 0.01), except SLS on both sides (p = 0.554 and 0.452). After 6 months of therapy, all parameters significantly improved compared to baseline. Specifically, gait velocity, step length and SLS of the more symptomatic knee improved by 13%, 7.8% and 3% respectively (p < 0.01). The changes in gait over the time intervals are summarised in Table 1. There was a significant improvement in KOFG between baseline and 3 months follow-up (p < 0.001), with retained improvements at 6 months. Table 2 (45). These stipulate either an improvement in total, pain or in function of at least 50% with a decrease of 2 cm on the visual analogue-scale for pain or function, or an improvement in both pain and function of at least 20% with a decrease of 1 cm on the VAS (45). 67% of the patients met the OMERACT-OARSI criteria (307/455 significantly improved). Table 3 displays the absolute changes from baseline to 6 months. All SF-36 subscales also significantly improved following 3 months of therapy (p < 0.001).
There were also further significant improvements between 3-and 6-month time intervals, except for the MCS (p = 0.068). After 6 months of therapy all subscales had significantly improved (p < 0.001). Specifically, SF-36 Total, PCS and MCS improved by 11.73, 15.7, and 9.62 points, or 22%, 34% and 15% respectively compared to baseline (See Table 3). These improvements also met minimal clinical important differences (MCID) for clinical significance of 7.8 points (46).
A correlation analysis between gait and questionnaire improvement was calculated. First,

Discussion
The purpose of this study was to evaluate the clinical effect of a personalised biomechanical foot-worn device on spatial-temporal gait parameters, pain, function and quality of life in UK patients suffering from knee OA. Following 3 months of therapy, patients demonstrated significant improvements in both gait and PROM's, with maintained or further improvements occurring in all parameters after 6 months. The results suggest that most improvements occurred by 3 months, but further improvements occur between 3 and 6 months, apart from SLS in gait and MCS subscale of SF-36. This supports previous work which found that the largest improvements occurred within the initial few months of therapy and are then maintained (32). The improvements in WOMAC scores meet the OMERACT-OARSI guidelines for clinical response to treatment, signifying true positive impacts felt by patients (45). The thresholds for minimal clinical importance differences (MCID) within SF-36 were also met, suggesting improvements in quality of life following treatment (46). In addition, these improvements in self-evaluation questionnaires correlated with the significant improvements in gait.
PROM's along-with radiographic findings have historically been used to track knee OA outcomes and are leading predictors in guiding the decision upon joint replacement surgery (47). However, the low correlation between radiographic findings and patient symptoms has since become more acknowledged (48), and therefore the importance of more objective and functional measures to evaluate patient symptoms has become apparent. Previous research has proposed spatio-temporal parameters are a good indicator for functional severity (12,48), with a recent meta-analysis suggesting that stride duration and cadence provided a better reflection of knee OA severity than kinematic and kinetic measures (49).
Using the patients' cadence and stride lengths as predictor variables for knee OA severity forms the basis of the validated Knee Osteoarthritis Functional Grade or KOFG (13), which in a follow-up study was also validated as a classification tool to measure treatment effect (50). A combination of spatial-temporal parameters objectively classifies patients with knee OA according to functional disease severity, which have been shown to correlate with radiographic evaluation, the level of pain, function and rate of TKR. The benefit of this tool is the ability to quantify the severity of disease and to assess the impact of an intervention, rather than just stating the change in gait analysis parameters (50). The model suggests that a shorter stride length with lower cadence is indicative of a higher functional severity grade (more severe knee OA), while a longer stride length with higher cadence is indicative of a lower functional severity grade (less severe knee OA). Within this current study, there was a significant improvement in KOFG between baseline and 3 months follow-up (p < 0.001), with retained improvement at 6 months. The shift from more severe to less severe KOFG suggest patients not only improvement in symptoms, but actually move away from measures associated with increased rates of TKR (13,50).
The present study also examined whether there were any differences in the sub-group of patients (20%) that had already been offered joint replacement surgery (TKR/PKR) prior to commencing the therapy. Interestingly, aside from the duration of symptoms, there were no significant differences between cohorts at baseline, suggesting that those patients that had been recommended surgery as a suitable option for their condition, displayed the same characteristics as those that have not had surgical recommendation (Fig. 3).
Research reports most patients that are suitable for TKR have WOMAC baseline scores between 40 s to 50's (47), which could indicate that the cohorts analysed within this study are representative of this population. Furthermore, significant improvements seen across both groups (Table 3)  patients access relevant treatments could be key to optimising outcomes (51). Despite this, the overall improvements in both groups are marked and suggestive that the therapy could be an effective alternative for a number of patients that would otherwise have progressed to surgery.
Researchers have presented several theories explaining how this therapy works to improve symptoms in patients with knee OA. Studies have shown that the biomechanical device can reduce the external loads acting on the body to "unload" the painful area, which is said to be important when treating this condition (40,41). It has been shown to reduce the 1st and 2nd peak KAM and KAAI by 8.4%, 12.7% and 13% after 9 months of therapy respectively (52). An important factor to note with the improvements in biomechanical variables seen with the device is that they remained, even when the patient was not wearing the device. This suggests that a motor learning effect occurs as a result of the neuromuscular retraining received from the therapy (43). Neuromuscular training is delivered by a controlled level of perturbation via the convex nature of the elements (43,44). The combined features of the biomechanical device allow for repetitive perturbations with diminished pain throughout the gait cycle. Patients wearing the devices for prescribed periods every day means that they gain high repetitions of closed kinetic chain, functional exercises and improved levels of compliance said to be advantageous for motor learning (53)(54)(55). This combination of key rehabilitation principles allows the patient to reacquire improved neuromuscular control, thus avoiding pathological patterns previously utilised whilst in pain (31).
Given knee OA is a chronic degenerative condition, symptoms deteriorate over time at varying rates (56). Currently, interventions are directed to the end-stages of the disease and therefore can often be ineffective and palliative in nature (57). Whilst the literature reports that the number of patients progressing to having joint replacement surgery is growing (6), approximately 20-40% of those completed are considered inappropriate.
These were classified due to having only slight or moderate symptoms, or not severe enough radiographic findings and therefore TKR deemed unnecessary (58,59). A paradigm shift is needed to focus efforts on treating patients at high-risk earlier in the disease progression (57), or utilising more specialist modalities that can help minimise this potentially inappropriate flow to surgery. The results of this study display a combination of improvements in both subjective PROM's and objective spatial-temporal gait parameters which are very promising and may indicate that the biomechanical device used in this study has the ability to be an effective modality in managing this chronic condition.
This study has some limitations. Firstly, the study was a retrospective analysis of patients from the centres database and therefore had no control group. However, a previous study has demonstrated comparable positive effects of this therapy compared to a control group in a double-blind study (31). Secondly, this study had a relatively short follow-up duration of 6 months for this cohort. Long-term follow-up would give more insight into the lasting effects of the treatment. However, it reflects previous research on the therapy on different populations with similar improvements in gait and PROM's (33)(34)(35)60).
Therefore, it could be assumed that the improvements can be maintained with the high compliance rates in the treatment (32). Nevertheless, future research should continue to investigate the long-term clinical effect of the treatment, in prospective, randomised control trial (RCT) design whilst tracking decay rates for joint replacement surgeries.
Promisingly, preliminary data from an RCT on the effect of this treatment displays comparable improvements to this study (61).

Conclusions
The examined non-invasive biomechanical treatment appears to provide an effective treatment that improves gait patterns, pain, function and quality of life for UK patients suffering with knee OA. It also appears to be equally effective on patients that have already been recommended knee joint replacement surgery, therefore potentially providing an alternative solution for this population. If these results can be retained in the longer term, it could hypothetically delay or even avoid the need for surgery in many cases. This would be positive for patients along with providing a more sustainable modality for healthcare systems to manage knee OA patients effectively in the community setting.

Declarations
Ethics approval and consent to participate The protocol was approved by the University of Roehampton Ethics Committee and all participants provided consent for their data to be used within in the study Consent for publication N/A Availability of data and materials All data request will be review and addressed by the authors

Competing interests
None to declare Flow chart of study screening and inclusion.

Figure 2
The biomechanical device Comparison of WOMAC changes between patients that had been recommended for knee joint replacement surgery (Total Knee Replacement/Partial Knee Replacement, TKR/PKR respectively) before commencing the biomechanical therapy and those that had not.