Study design and participants

This study was designed as a randomised controlled trial with two independent groups and an allocation ratio of one to one. The responsible federal ethics committee Kantonale Ethikkommission Bern stated, after reviewing, that the study did not fall under the Swiss Human Research Act and did not require further approval (BASEC-Nr.: Req-2021-00554). Trial registration was not applicable due to the cross-sectional design (one measurement point, no intervention).

An a priori sample size was calculated for two independent groups (significance level = 0.05; power = 0.80) and resulted in N = 9 per group [19]. The effect size was based on the results of a similar study with the used mean difference as follows: intervention group mean ± standard deviation (SD) 199 ± 39 N, control group mean 251 N [20]. Participants were recruited by hanging up flyers and sending emails to organisations and healthcare clinics within the local community. Inclusion criteria were being a minimum of 60 years of age, having a European shoe size of 36–45 and being able to navigate stairs and be able to walk without a walking aid for 10 min. Exclusion criteria were the presence of any health issue impacting their gait or ability to use crutches, severe cognitive impairment, and having used crutches within the last six months.

Data collection

Measurements took place from November 2021 to March 2022 in the Bern Movement Lab at the Bern University of Applied Sciences or in the  SportClinic facilities in Zürich, Switzerland. Volunteers were informed of the study procedures and, if they provided written consent, demographic data were recorded. Grip strength was measured as a quick, valid measure of upper body strength using a hydraulic handgrip dynamometer (JamarFootnote 1) following the Southampton protocol for adult grip strength measurement [21, 22]. Because cognitive impairment may contribute to poorer compliance in older adults [15, 23], level of cognitive function was evaluated using the German version of the Montreal Cognitive Assessment (MoCA). This validated test is best suited for quickly detecting mild cognitive impairment among people older than 60 years [24, 25]. Levels were defined according to test developers’ recommendations, with no cognitive impairment a score of > 25, mild a score of 18–25, moderate a score of 17–10, and severe a score of < 10 [26].

The primary outcome measure was the amount of weight-bearing on the PWB leg during six functional mobility activities, measured as the ground reaction force in Newtons and converted to kilograms to improve clinical interpretation. Sensor insoles were used to record the ground reaction force (OpenGo, Insole3Footnote 2). The third generation insole3® model is a valid and reliable instrument for measuring vertical ground reaction forces during walking, according to validation studies independent of the producer [27, 28].

Protocol

Sensor insoles were fitted to the participant and calibrated. The non- (and later partial-) weight-bearing leg was randomly determined using a mobile coin toss application (Tiny Decisions App, Version 2.9.1.Footnote 3). Elbow crutches were fitted and a swing-through non-weight-bearing gait instructed, with the participant given as much time as they needed to feel comfortable and safe [29]. Next, the volunteer was allocated to either the intervention protocol or the control protocol using an online randomisation service, that randomly assigned the individual participant to one of the two groups (Sealed EnvelopeFootnote 4). Blinding of intervention allocation was not possible. The PWB training protocol was based on the authors’ knowledge of clinical practice and the methodology of similar studies [12, 30,31,32]. The same experienced and licenced physiotherapist instructed all participants to PWB to a limit of 20 kg according to the protocol of the group to which they were allocated to. No changes were made to the eligibility criteria nor the protocol after commencement of the trial.

In the control group, the target PWB load was instructed using an analogue scale (ADE M308800Footnote 5). The participant stood with their PWB leg on the scale and transferred weight onto this leg until the needle pointed to 20 kg on the dial. They were instructed to pause a moment and try to remember how this degree of load felt before lifting the foot off the scale. They repeated this process five times. Next, the participant stood with the same leg on the scale but looked straight ahead. They were instructed to transfer weight onto this leg until they felt they were loading it to 20 kg, then to look at the dial and correct the pressure, if necessary, before removing the foot off the scale. They repeated this process a total of five times.

For the intervention group, the OpenGo system was used to provide audio-biofeedback using the OpenGo App installed on a smartphone. Using the mobile application’s inbuilt function, 200 N (approximately 20 kg) was manually set as the threshold load for the PWB leg. The participant was informed that, when this function was activated, the application would notify them with a beep when the threshold load of 20 kg had been reached and would continue to beep as long as the load was at or above this threshold. The participant was instructed to place their PWB foot on the floor, transfer weight onto the leg until they heard a beep. Then, they were told to adjust the pressure and, by listening to the biofeedback, learn what amount of pressure constituted 20 kg of load, before lifting their foot off the floor again. The participant repeated this procedure 10 times.

Following, the participant was instructed to walk with a three-point PWB gait and told to try to weight-bear as close as possible too, but not over, 20 kg. They were instructed how to use crutches to stand up from and sit down on a chair and step up onto and step down from a step. In the intervention group, the audio-biofeedback remained activated during this part of training, meaning they had concurrent auditory feedback set to 200 N, whereas the scale group did not. Upon completion of the training session, the audio-biofeedback function was turned off and no further feedback on the amount of weight-bearing was provided to participants in either group. Participants were allowed to practice these activities until they and the physiotherapist were confident that they could perform the activities safely. The training time required varied from approximately 10 to 20 min.

The amount of weight-bearing was recorded at 100 Hz during six functional mobility activities:

1.

standing for 30 s,

2.

standing up from, and sitting back down on, a chair,

3.

stepping up onto a single step,

4.

stepping down from a single step,

5.

walking on a flat surface for three minutes, and

6.

walking with a 4 kg weighted backpack for one minute.

The sit-stand-sit and step activities were measured three times. Prior to each activity, the participant was reminded they should weight-bear close to, but not over, 20 kg.

Data processing and analysis

The recorded de-identified data was transmitted to the OpenGo Desktop Software (Version 2.1 [see also 2]) and processed using the inbuilt analysis functions. Recordings were deleted from the insole memory immediately after transfer. For the walking activities, the mean of all maxima of total force during the stance phase was reported. For the sit-stand-sit and stepping activities, the maximum total ground reaction force was recorded. For standing, the mean total force during the first 30 s of the recording was used instead, because standing is considered a static activity where the load over time is more relevant than the maximum load. All demographic and sensor insole data were entered into an Excel® spreadsheet on a password-protected computer and hardcopies stored in a locked cabinet.

Statistical analyses were performed using R (Version 4.1.2Footnote 6). Means and standard deviation of PWB were calculated for each group for all activities and because data were not normally distributed (as tested by the Shapiro–Wilk Test) the Mann–Whitney U Test was used to compare the differences between the two groups. Individual and group adherence to weight-bearing was evaluated by comparing PWB means against the 20 kg target load within a pre-defined 10 kg buffer zone (upper limit: 25 kg, lower limit: 15 kg). The selected adherence range was comparable to those used by similar studies [33, 34]. For each participant, the number of steps within and outside this adherence range was manually counted and step percentages for each load classification (adherent, non-adherent overload and non-adherent underload) calculated. The relationship between each independent factor (age, gender, body mass index (BMI), grip strength and MoCA score) and the dependent variable mean PWB load (kg), was assessed via simple linear regression. Finally, multiple linear regression analysis was performed using background knowledge to guide the development of three different models, for each model, the Akaike information criterion was calculated to select the best-fit model across all activities [35]. The confidence interval (CI) was set at 95% and a p-value of 0.05 was defined as statistically significant.