This study is an observational cross-sectional study. Ethical approval was obtained from the University Ethics Committee in September 2021 (CERUBFC-2021-11-09-036). The study was performed in accordance with the ethical practices outlined in the Declaration of Helsinki (1964).

Fifty-one volunteer older adults (76.9 ± 7.6 years, 13 males and 37 females) were recruited by a physiotherapist from a readaptation center located in Montbéliard (France) between February 2022 and June 2023.

All the participants were aged 65 years and over and presented with some orthopedic disabilities (hip or knee arthroplasty). However, all participants presented a corrected vision deficit and an ankle muscle strength > 3/5 on manual muscle strength assessment. None of them had diabetes, vestibular pathology or major neurocognitive disorders. Written informed consent was obtained from each participant.

The participants’ anthropometric data (gender, age, and body mass index) were collected. Fall history from the previous 6 months and the presence (or absence) of fear of falling (assessed using the Short-FES scale) [17] and neck pain (visual analog scale) were also recorded. The range of ankle dorsiflexion was measured for each participant, as were their grip strength and gait speed. Ankle dorsiflexion range of motion was measured bilaterally using a inclinometer [18]. Grip strength was assessed using a Jamar® dynamometer, following established guidelines [19]. Gait speed was assessed over a 10-meter distance, with time measured at self-selected pace [20].

The participants were divided into two functional groups (FG-/FG+) according to a composite score that included 3 variables (gait speed, grip strength and fear of falling).

The participant’s gait speed was expressed relative to the maximum gait speed of our sample (m.s− 1).

GSpeed’ = (Gait speed × 100)/1.2.

The participant’s grip strength was expressed relative to the maximum grip strength of our sample (kg).

GStrength’: (Grip strength × 100)/46.9.

The composite score was subsequently calculated via the following formula:

Composite score = (Gspeed’+Gstrength’) +/- 10%.

+/- 10% was applied depending on the presence (+) or absence (-) of fear of falling.

The protocol included an evaluation session performed by an evaluator unaware of the participants. All the participants completed the modified Clinical Test of Sensory Interaction on Balance (m-CTSIB) and performed one proprioception test. The exclusion criterion encompassed incomplete stabilometric evaluations, defined as participants completing only 2 out of 3 trials per condition or evaluating only 2 out of 4 specified conditions.

The test used a stabilometric platform (Techno concept® with Posturewin software, France). The participants were instructed to stand barefoot on the platform and adopt the reference position: feet shoulder width apart, arms at their sides and gazing straight ahead at a visual cue. Throughout the 20-second data acquisition, the participants maintained stillness while their center-of-pressure (CoP) trajectory was captured at a rate of 40 Hz by the platform’s three strain gauges. The plate level thickness was 0.002 m from the ground. A foam surface (Airex®, height 50 cm, width 41 cm, thickness 6 cm, density 55 kg.m− 3) was used to disrupt the contribution of podal information to postural regulation.

Four conditions, comprising a measurement block, were assessed: standing reference position with eyes open (condition 1), standing with eyes closed (condition 2), standing on foam with eyes open (condition 3), and standing on foam with eyes closed (condition 4). The participants completed three measurement blocks, separated by breaks between each block. If participants required assistance to maintain balance, e.g., with light support on a chair or the physiotherapist to prevent them from falling, the trial was stopped and not retained.

After executing the m-CTSIB test, the participants were instructed to perform the (neutral) cervical repositioning test (or cervical joint sense position error) [21]. In the sitting position, participants wore a helmet with a laser pointer and were asked to memorize the neutral reference position, which consists of looking straight ahead with the laser pointing at the center of a target 90 cm away. With the eyes closed, the participants performed active rotational movements in the transverse plane through the full range of motion, from left to right. The objective was to return to the reference position with maximum precision consistently with the eyes closed. Each trial involved measuring the distance from the target’s center to the laser’s arrival point, which was determined when the participant believed that he or she had accurately returned to the reference position.

Three trials were performed, and the results were averaged. In accordance with the same principle, three cervical repositioning tests were also performed after active flexion/extension movement (sagittal plane) [22].

For the cervical repositioning test, the difference between the starting position (zero) and the point of return in the plane of movement was measured in centimeters and then converted to degrees via the following formula: angle = tan− 1 [error distance/90 cm] [23]. Only the absolute error was calculated and defined as the mean of the total deviation from the target, ignoring the positive and negative values [24].

Before conducting the statistical analysis, the normality (Shapiro‒Wilk test, all p > 0.01) of each variable was checked. The comparison of fundamental clinical data (shoe size, weight, body mass index, dorsiflexion, performance of cervical proprioception tests) and stabilometric data (CoP mean velocity during conditions 1, 2 and 3 of the m-CTSIB) between the two groups utilized the Mann‒Whitney test. Student’s t test was used to compare the following variables: age, height, walking speed and CoP mean velocity during condition 4 of the m-CTSIB. Additionally, a χ² test with Yates correction was performed for the variable “history of falls”.

An exploratory factor analysis based on a matrix of correlated associations was carried out to determine common factors between variables. Considering the nonnormality of several variables, the method used for factor extraction was the “principal axis factor” procedure. Next, Bartlett’s test was used to verify the hypothesis of sphericity, and the Kaiser‒Meyer‒Olkin (KMO) test was used to check the adequacy of the sampling. The number of factors was determined via the parallel analysis method, which considers the number of factors whose eigenvalues are greater than those obtained with random data. The sum of the square loadings and the proportion of variance were subsequently calculated. The applied rotation method was either oblique or orthogonal, depending on the factor loading coefficient.

If stabilometric variables seem to be linked with repositioning variables, and in the case of normality and homogeneity of these variables, a characterization of the relationship was performed via the Pearson coefficient.

For all the statistical analyses, the significance level was set at p < 0.05. Statistical analyses were performed via JASP® software (Version 0.16.3, JASP Team, University of Amsterdam, Netherlands).