Inter-rater reliability of the instrumented lean-and-release (iLEAN) test for manual assessment of reactive balance: a pilot study in simulated older adults

Abstract

[Purpose] Quantitative assessment of reactive balance often requires large, costly equipment, limiting its use in clinical practice. This study investigated the inter-rater reliability of the instrumented lean-and-release (iLEAN) test, a manual reactive balance assessment designed to provide a practical alternative for physiotherapists. [Participants and Methods] Sixteen healthy young women completed the iLEAN test in standing, where reactive balance control was assessed in forward, left, right, and backward directions. During tests, an assessor applied the iLEAN device against the shoulder and released support once the participant reached a target lean load, which was progressively increased until balance could not be regained with a single step. Inter-rater reliability was analyzed using intraclass correlation coefficients (ICC, model 2,1) and Bland–Altman plots. [Results] The ICCs (95% confidence intervals) for the iLEAN scores were 0.60 (0.17–0.84) for forward, 0.75 (0.34–0.92) for left, 0.80 (0.45–0.94) for right, and 0.83 (0.58–0.94) for backward. Bland–Altman analysis revealed no systematic bias between raters across all directions. [Conclusion] This study provides preliminary evidence that the iLEAN test has acceptable inter-rater reliability for assessing reactive balance, with the backward condition demonstrating the highest reliability and sensitivity to change.

INTRODUCTION

Falls in community-dwelling older adults are often precipitated by postural perturbations, such as trips and slips during walking¹⁾. Reactive balance—the ability to restore posture following a sudden mechanical disturbance—is a critical determinant of fall prevention²⁾. Evidence from systematic reviews indicates that older adults with a history of falls exhibit impaired reactive stepping responses compared to their non-falling peers³⁾.

In clinical practice, reactive balance is typically assessed with the pull test or the push-and-release test. These methods are low-cost and have demonstrated good reliability in populations such as people with Parkinson’s disease and polymyositis⁴⁾. However, their precision is limited due to variability in perturbation delivery and the qualitative nature of response scoring. Conversely, the lean-and-release test offers a more standardised and quantifiable assessment of reactive stepping⁵⁾, and it has been reported that individuals who required multiple recovery steps following 15–20% body-weight perturbation loads were at a significantly greater risk of future falls⁶⁾. However, its implementation in routine practice is hindered by the need for specialised equipment, large space, and technical expertise.

To address this gap, we developed the instrumented lean-and-release (iLEAN) test, which incorporates two hand-held dynamometers to provide a portable, manual, and quantitative assessment of reactive stepping in forward, backward, and lateral directions. The purpose of this preliminary study was to evaluate the inter-rater reliability of the iLEAN test in healthy young adults under simulated age-related physical function constraints. Healthy young participants were tested under simulated aging conditions to reduce safety risks during this pilot phase, as the protocol involved progressively increasing leaning loads without a safety harness.

PARTICIPANTS AND METHODS

Participants were recruited through university bulletin board advertisements. Inclusion criteria were healthy women without neurological, musculoskeletal, or psychiatric disorders. The study was approved by the Office of Research Ethics at Kanagawa University of Human Services (Approval No. 18-23-34) and complied with the Declaration of Helsinki. All participants provided written informed consent.

Baseline measures included age, sex, height, weight, body mass index, and dominant leg. The stepping leg in the forward and backward directions was defined as the dominant leg, which was identified by asking: “Which leg would you use to kick a ball?”⁷⁾.

Reactive balance was assessed in forward, backward, left, and right directions using the instrumented lean-and-release (iLEAN) test. The device comprised two hand-held dynamometers (μTas, Anima Corp., Tokyo, Japan) held by the assessor and a wrist-worn display connected to an Android device running the iLEAN application (Fig. 1). For the forward, laterals, and backward trials, the hand-held dynamometers were placed to the glenohumeral joint, acromioclavicular joint, and inferior angle of the scapula, respectively.

Fig. 1.

iLEAN device and application.

To simulate age-related functional decline, participants wore a 2.7-kg weighted vest, 1.5-kg ankle weights, and bilateral knee braces (Oitarou, Kyoto Kagaku, Kyoto, Japan). For forward and backward trials, participants stood with feet hip-width apart; for lateral trials, feet were together. A side wall was positioned 50 cm away for safety. In forward, left, and right conditions, a foam block (14 cm height, placed 5 cm in front of the stepping foot) was used to create an earlier failure point, reducing the need for excessive leaning angles and ensuring participant safety during the test. The foam block also encouraged a more elevated stepping response which is a key characteristic of compensatory stepping following a trip⁸⁾.

For each trial (Fig. 2), participants folded their arms across the chest, leaned against the device until a target load was reached, and were then released by the assessor within 1–5 seconds. Participants were instructed to evenly distribute their body weight between both lower limbs when assuming the initial leaning posture. The initial load was set at 5% of body weight and increased in 1% increments if balance was recovered with a single step. The assessors provided clear instructions on how to conduct the test as follows: “For this test, I would like you to lean against my hands to reach a target load. Your first target is 5% of your body weight which is XX kg. When leaning, make sure to keep your body straight and relax your ankles. You will hear a beep when you reach the target load. Once this happens, I will quickly release my support within a 1–5 second interval. From here, you will take only one step with your XX leg to regain balance. If you succeed, I will continue to increase your target load until you fail. In forward and backward trials, you should take only a single step with your dominant leg to regain posture. In lateral trials, you should not perform a cross-over step to regain posture”. Before the test, participants conducted one practice trial without/with a light load to familiarise the participant with the task and ensure they use the specified leg. The maximum load of 20% of body weight was set as a safety limit because the risk of falling and the assessor burden increase substantially beyond this threshold. A trial was considered unsuccessful if participants: (1) unfolded arms, (2) touched the wall, (3) took more than one step, (4) touched the floor with a knee, or (5) in forward/lateral directions, displaced the foam block. In such cases, participants were given one additional attempt. If balance was successfully recovered in either the first or second trial, the target load was increased by 1%. The highest target load successfully recovered was recorded as the iLEAN score (% body weight).

Fig. 2.

Scenes of the iLEAN test.

a), b), c) and d) show the scenes of the tests in the forward, left, right and backward directions, respectively.

Two trained physical therapy students (A.N. and Y.K.) served as assessors. We selected the physical therapy students to assess whether reproducible measurements could be obtained even by those with limited experience and immature skills. Order of assessors was randomized using the envelop method, and assessors were blinded to each other’s results by being kept away from the measurement room.

Inter-rater reliability for iLEAN scores in each direction was evaluated using intraclass correlation coefficients (ICC, model 2,1) with 95% confidence intervals (CI). ICCs were interpreted as: <0.50=poor, 0.50–0.75=moderate, 0.75–0.90=good, and >0.90=excellent⁹⁾. Bland–Altman plots were used to assess systematic bias. Fixed bias was identified when the 95% CI of mean differences did not include zero, and proportional bias was determined from significant correlations between mean scores and differences. Minimal detectable change at 95% confidence (MDC₉₅) was calculated as: SEM ×√2 × 1.96¹⁰⁾.

Sample size requirements were estimated using Walter et al’s¹¹⁾ method, indicating that 12 participants would be sufficient to detect an ICC of 0.9 with a lower CI of 0.60 (α=0.05, β=0.20). To account for potential dropouts and outliers, we recruited 16 participants. Statistical analyses were conducted using SPSS 26.0 for Windows (IBM Corp., Armonk, NY, USA).

RESULTS

Sixteen healthy women (average age 20.7 years, standard deviation 0.8 years) were included in this study (Fig. 3, Table 1).

Fig. 3.

Flow chart of participants included in this study.

Table 1. Participant characteristics.

Items
Age (years)	20.7 ± 0.8
Height (cm)	158.5 ± 5.3
Weight (kg)	52.4 ± 7.4
BMI (kg/m2)	20.8 ± 2.0
Dominant leg (right), n (%)	16 (100)

Data are presented as mean ± standard deviation or number (percentage). BMI: body mass index.

Table 2 presents the iLEAN scores and ICCs for each direction. Scores could not be obtained for four participants in the left and right conditions because they were unable to complete the trials at the initial target load without striking the foam block by constrained foot elevation. The ICC (2,1) values were 0.60 (95% CI: 0.17–0.84) for forward, 0.75 (95% CI: 0.34–0.92) for left, 0.80 (95% CI: 0.45–0.94) for right, and 0.83 (95% CI: 0.58–0.94) for backward directions.

Table 2. iLEAN scores and inter-rater reliability.

Directions	iLEAN scores (%)		ICC (2,1)	95% CI of ICC (2,1)
	Rater A	Rater B
Forward (n=16)	12.4 ± 3.0	12.6 ± 3.4	0.60	0.17 to 0.84
Left (n=12)	7.5 ± 2.5	7.8 ± 2.1	0.75	0.34 to 0.92
Right (n=12)	8.4 ± 2.2	8.7 ± 2.4	0.80	0.45 to 0.94
Backward (n=16)	9.3 ± 2.1	9.8 ± 1.9	0.83	0.58 to 0.94

Data are presented as mean ± standard deviation. ICC: intraclass correlation coefficients; CI: confidence interval.

Bland–Altman plots (Table 3, Fig. 4) showed that most data points lay within the limits of agreement (± 1.96 SD of the differences between raters), with no evidence of fixed or proportional bias across directions. The MDC₉₅ values were 2.07 (forward), 1.42 (left), 1.37 (right), and 0.85 (backward).

Table 3. Bland–Altman analysis.

Directions	Fixed bias		Proportional bias			MDC95
	95% CI	(+) (−)	Regression	p-value	(+) (−)
Forward (n=16)	−1.77 to 1.25	(−)	−0.16	0.545	(−)	2.07
Left (n=12)	−1.23 to 0.83	(−)	0.22	0.494	(−)	1.42
Right (n=12)	−1.24 to 0.63	(−)	−0.14	0.660	(−)	1.37
Backward (n=16)	−1.07 to 0.16	(−)	0.14	0.604	(−)	0.85

CI: confidence interval; MDC: minimal detectable change.

Fig. 4.

Bland–Altman plot of iLEAN tests.

Vertical axis is the difference of each value between two assessors for each participant and horizontal axis is the average of each measurement of two assessors. Dotted line shows the mean of differences of the values of two assessors ± 1.96×SD of the differences of the values of two assessors.

DISCUSSION

This pilot study evaluated the inter-rater reliability of the iLEAN test for assessing reactive stepping ability in healthy young adults. All directions demonstrated acceptable reliability, with the backward condition showing the highest ICC and the smallest MDC₉₅ (0.85). Clinically, this suggests that even small changes in backward iLEAN scores may represent meaningful improvements in reactive balance beyond measurement error.

The backward iLEAN test compared favorably with existing manual assessments of reactive balance. Its ICC (0.83) was higher than the pull test (ICC: 0.45 on the first trial and 0.74 on the third trial)⁴⁾ and comparable to the push-and-release test (ICC: 0.84 on the first trial and 0.83 on the third trial)⁴⁾. This may reflect the more controlled force application of iLEAN and push-and-release, in contrast to the variable abruptness of the pull test. Importantly, unlike the push-and-release test, which uses ordinal ratings (a 5-point scale from 0 to 4), the iLEAN provides a quantitative measure of reactive stepping capacity, potentially reducing floor and ceiling effects and allowing more sensitive monitoring of change.

In terms of the iLEAN tests in the other directions, the inter-rater reliability in the forward direction was lower (ICC: 0.60, 95% CI: 0.17–0.84) than those in other directions. This may be due to the greater leaning loads that assessors were required to support in the forward direction (12.4 ± 3.0%) than other directions (e.g. left: 7.5 ± 2.5%) which may contribute to fatigue and variable holding performance of the assessor during the test. Although the ICCs for the iLEAN tests in the left and right directions showed good inter-rater reliability (ICC: 0.75 for the left direction and 0.80 for the right direction), four participants were unable to complete the initial 5% body weight trials in both the left and right directions due to their stepping foot contacting the foam obstacle. This suggests the obstacle may not be necessary, particularly in clinical populations where balance recovery is likely to be impaired.

Several limitations should be acknowledged. First, despite wearing the geriatric simulation suit mechanically limiting the joint motion¹²⁾, the findings from young healthy women may not fully generalize to older adults or clinical populations. Additionally, using constant weights across participants, despite differences in body size, may have influenced the degree of simulated impairment. Second, the initial target load of 5% body weight may be too high for frailer individuals, and lower starting loads (e.g., 3%) should be considered. Although a foam block was used to create an earlier failure point—reducing the need to manage excessive leaning angles and associated safety risks, as well as facilitating a more elevated stepping response similar to trip recovery—the use of such a constraint is debatable. Omitting the foam block may allow for more natural stepping and improve practicality, particularly in older populations. Third, test–retest reliability was not examined. Fourth, only women participated, and further studies should include men with a range of body sizes. As the iLEAN test applies force at shoulder height, proportional body segment ratios (e.g., shoulder and center of mass height) could be relatively consistent across individuals, meaning that an identical % body weight force likely produces similar perturbation effects regardless of stature. Future work could confirm this assumption using detailed anthropometric data and by comparing results with conventional lean-and-release method (e.g., maximum recoverable lean magnitude or body size). Finally, in this study, stepping responses in forward and backward directions were assessed only with the dominant leg; however, in older adults or people with unilateral impairments (e.g., stroke, Parkinson’s disease), both legs may need to be tested because lateral differences may be expected to be significant.

This study provides preliminary evidence that the iLEAN test has acceptable inter-rater reliability for assessing reactive balance, with the backward condition demonstrating the highest reliability and sensitivity to change. The iLEAN test may therefore offer physiotherapists a practical, quantitative method for monitoring improvements in reactive balance during exercise or rehabilitation without reliance on costly, laboratory-based equipment. Future research should confirm these findings in older adults and patient populations, and further evaluate intra-rater and test–retest reliability with a large sample size.

Preprint publication

Hirase T, Nishimura A, Kato Y, et al. Inter-rater reliability of the instrumented lean-and-release (iLEAN) test for manual assessment of reactive balance: a pilot study in simulated older adults. medRxiv, Preprint posted online September 12, 2025. Doi: 10.1101/2025.09.10.25335475

Conflict of interest

The authors declare no conflicts of interest.