The aging population is a demographic trend observed in more than 100 countries worldwide. By 2050, more than 30% of nations are expected to transition into a super-aged society, where more than 20% of the population is 65 years or older [1]. Aging in older adults is associated with physiological changes, such as cognitive decline, muscle mass reduction, and decreased bone density, which increase the likelihood of injury during falls [2]. Falls are most commonly linked to severe injuries, such as hip fractures and brain injuries, which often lead to prolonged disability [3].

Aging accelerates significantly after 50 years of age [4], particularly affecting proprioception, which plays a critical role in maintaining balance and coordinating movement [5]. Proprioception is a sensory system that enables the body to perceive its position and movement without relying on visual cues, functioning through afferent signals from sensory receptors in joints, muscles, and tendons [6]. A decline in proprioceptive signals from the lower limb receptors can impair balance control [7]. Furthermore, as proprioception in the ankle joint deteriorates, the hip joint compensates by taking on a greater role [8]. This shift means that, with aging, the delayed neural transmission and weakened joint position sense reduce the ability to recover balance or adjust posture swiftly, increasing the risk of falls [9].

Studies have shown that proprioceptive training can enhance proprioception in individuals with knee osteoarthritis, chronic sprains, Parkinson’s disease, and stroke [10]. These findings suggest that even impaired proprioception can be improved through targeted interventions. The use of reliable and valid evaluation tools is crucial for assessing these improvements. Currently, proprioception is assessed using methods such as joint position sense (JPS), which measures the awareness of joint positioning; threshold to detection of passive motion (TTDPM), which evaluates the sensitivity to subtle joint movements; and force sense (FS), which assesses the perception English of applied force [11-13].

On the other hand, a comprehensive evaluation of the reliability of these methods, particularly in assessing lower limb proprioception, is limited. Previous systematic reviews focused primarily on upper limb proprioception [14,15] or aspects, such as pain [16]; and motor control [17], leaving a significant gap in the literature concerning the reliability of lower limb proprioception measurement tools. This lack of research addressing the reliability of proprioception assessments in the hip, knee, and ankle joints presents a challenge for clinicians aiming to evaluate and improve proprioceptive function in these areas.

Therefore, this study reviewed the reliability of proprioception measurement tools for the hip, knee, and ankle joints in adults aged 50 and older. This research provides a foundation for developing more effective and standardized methods for assessing lower limb proprioception in clinical settings by evaluating clinically applicable tools.

1. Study Design

This study was a systematic review conducted in accordance with the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines [17], aiming to evaluate the reliability of various methods for measuring proprioception in the lower limb joints of adults aged 50 years and older.

2. Search Strategy and Study Selection

A literature search was conducted using PubMed, Excerpta Medica Database (EMBASE), Cochrane Central Register of Controlled Trials (CENTRAL), and Web of Science in July and August 2024. The search strategy included the keywords “elderly” AND “proprioception” AND “reliability” (Table 1). The search concluded on August 15, 2024.

Table 1

Search strategy used in the PubMed, EMBASE, CENTRAL, and Web of Science databases

Keyword	Search Terms
Elderly	"Elderly" OR "Older adults" OR "Seniors" OR "Aged individuals" OR "Older people" OR "Elderly population" OR "Geriatric population"
Proprioception	"Proprioception" OR "Proprioceptive sense" OR "Proprioceptive acuity" OR "Proprioceptive accuracy" OR "Proprioceptive sensitivity" OR "Kinesthesia" OR "Proprioceptive feedback" OR "Proprioceptive assessment" OR "Joint Position Sense Test" OR "joint position reproduction" OR "active joint position sense" OR "threshold to detection of passive motion" OR "TTDPM" OR "motion detection threshold" OR "active movement extent discrimination apparatus" OR "AMEDA"
Reliability	"Reliability" OR "Test-retest reliability" OR "Intra-rater reliability" OR "Inter-rater reliability" OR "Validity" OR "Usability" OR "Safety" OR "Assessment consistency" OR "Clinical feasibility" OR "Measurement precision"

AMEDA = Active Movement Extent Discrimination Assessment, TTDPM = Threshold to Detection of Passive Movement

The studies identified were imported into RefWorks, where duplicates were removed based on title, author, and publication year. Two independent reviewers screened the studies based on predefined inclusion and exclusion criteria. Initial screening was conducted by reviewing titles and abstracts, followed by a full-text review to finalize the study selection. In cases of disagreement between the two reviewers, discussions were held to reach a consensus, and the corresponding author acted as a mediator when necessary.

1) Inclusion Criteria:

⦁ Studies involving adults aged 50 and older.

⦁ Studies providing separate data analysis for participants aged 50 years and older.

⦁ Studies assessing the reliability of lower limb proprioception measurements.

⦁ Published in English.

2) Exclusion Criteria:

⦁ Studies that only used questionnaires for proprioception measurement.

⦁ Studies evaluating proprioception of joints other than the lower limbs.

⦁ Studies involving only diseased adults aged 50 and older.

3. Data Extraction

The selected studies were categorized using Microsoft Excel (Office 365, Microsoft Corporation, USA) as follows: proprioception outcome, participant characteristics (number, sex, age, and status), assessor characteristics (number, experience, and background), intraclass correlation coefficient (ICC) model and type, test-retest period, proprioception equipment, and body part involved in proprioception measurement.

4. Quality Assessment

Two independent reviewers assessed the methodological quality of the studies using the COSMIN Risk of Bias tool [18]. The standards for reliability and measurement error studies were applied to evaluate the reliability of the outcome measurement instruments and the associated measurement error.

The COSMIN Risk of Bias tool evaluates studies on a four-point scale of "very good (4)," "adequate (3)," "doubtful (2)," and "inadequate (1)." Each item is rated individually, and no composite score is calculated. On the other hand, item 6 follows the "worst score counts" principle, which evaluates whether the study has any significant design or statistical flaws. The study quality may be questioned if rated "doubtful" or "inadequate." The standards for reliability studies include the following nine items: 1) stability of the patient, 2) time interval, 3) similar measurement conditions, 4) administration of measurements, 5) assignment of the score, 6) design or statistically important flaws, 7) preferred statistical methods for continuous scores in reliability, 8) preferred statistical methods for ordinal scores, and 9) preferred statistical methods for dichotomous or nominal scores.

The standards for the measurement error studies included the same six items as the reliability studies and an additional two: 7) preferred statistical methods for continuous scores in measurement error and 8) preferred statistical methods for dichotomous, nominal, or ordinal scores in measurement error.

1. Selection and Characteristics of the Studies

Three hundred and forty-four studies were identified through the literature search. Sixteen duplicate studies were removed based on the title, author, and publication year. An additional 306 studies were excluded after screening the titles and abstracts. Five studies were excluded because of inaccessible full texts, and twelve were excluded because they did not align with the objectives of this systematic review. Ultimately, 339 studies were excluded, and five studies that assessed the reliability and validity of various proprioception measurement methods for lower limb joints in adults aged 50 and older were selected for inclusion (Fig. 1).

Fig. 1. PRISMA flowchart.

Four hundred and fifty-six participants were included across the five studies, most of whom were healthy young adults or cognitively healthy older adults. The participants' ages ranged from 25 to over 85 years. Three studies used the ICC(2,1) model, while one study used the ICC(2,k) model (Table 2).

Table 2

Characteristics of the studies

Study	Participant characteristics				ICC model & type
	(N)	Sex(M/F)	Age	Health status
Antcliff et al. 2021 [22]	44	17/27	20–64 years: n = 29 65–84 years: n = 11 85+ years: n = 4	Healthy adults and older adults	(2,1)
Arvin et al. 2015 [12]	19	5/14	73.5 years	Healthy older adults	(2,k)
Baert et al. 2018 [11]	34	KOA = 4/4 without KOA = 9/17	KOA: Mean age 68 years without KOA: Mean age 58 years	KOA and Healthy older adults	(2,1)
Ko et al. 2015 [13]	289	158/131	Males: 73 years Females: 70 years	Cognitively intact older adults	NS
Westlake et al. 2007 [26]	70	Healthy older adults = 27/19 Younger adults = 6/18	Healthy older adults: Mean age 76.28 years Younger adults: Mean age 26.17 years	Healthy adults and older adults	(2,1)

ICC = Intraclass Correlation Coefficient, KOA = Knee Osteoarthritis, NS = Not stated

2. COSMIN Risk of Bias in the Studies

In this review, no studies applied to sections 8 and 9 of the standards for studies on reliability. Therefore, the evaluation was limited to items 1 to 7 from the reliability standards and items 7 and 8 from the standards for studies on measurement error, resulting in nine items. For item 1, one study was rated “very good,” while four were rated “adequate.” For item 2, three studies were rated “very good,” and two were rated “adequate.” All five studies received a “very good” rating for item 3. In item 4, one study was rated “very good,” two “adequate,” and two “doubtful.” For item 5, two studies were rated “very good,” one “adequate,” and two “doubtful.” For item 6, three studies were rated “very good,” one “adequate,” and one “doubtful.” All studies were rated “very good” for item 7. For item 8, two studies received a “very good” rating, and for item 9, one study was rated “very good” (Table 3).

Table 3

Risk of Bias in the Studies

Study	Stability of the patient	Time interval	Similar measurement conditions	Administration of measurements	Assignment of the score	design or statistically important flaws	Preferred statistical methods for reliability	Preferred statistical methods for measurement error	Preferred statistical methods for scores
Antcliff et al. 2021 [22]	3	4	4	2	2	2	4	-	-
Arvin et al. 2015 [12]	3	4	4	2	2	3	4	4	-
Baert et al. 2018 [11]	3	3	4	3	4	4	4	4	4
Ko et al. 2015 [13]	3	3	4	3	3	4	4	-	-
Westlake et al. 2007 [26]	4	4	4	4	4	4	4	-	-

3. Summary of the Reliability Data from the Included Studies

The reliability of proprioception in older adults was assessed across one study for the hip joint, two for the knee joint, and three for the ankle joint. The proprioception measurement methods included JPS, active movement extent discrimination apparatus (AMEDA), knee force sense (KFS), TTDPM, and threshold-to-velocity discrimination (TTVD). The equipment used included a 3D Motion-Capture System, an analogue inclinometer, a handheld dynamometer, a platform, and an isokinetic dynamometer. Only two studies mentioned the number of assessors, and only one described their characteristics. All five studies used the ICC to evaluate the reliability (Table 4).

Table 4

Reliability Data of the Studies

Body part	Study	Proprioception outcome	Equipment	Repetitions and test-retest period	No. assessor and experience and background	Reliability(ICC)
Hip	Arvin et al. 2015 [12]	Active JPS: 10-45° Flexion	3D Motion-Capture System	4 trials on each limb and two sessions (within 24 hours)	1 and NS	RE: .77-.79 AE: .11-.56 VE: .06-.59
		Active JPS: 10-40°Abduction				RE: .81-.93 AE: .71-.81 VE: .36-.70
Knee	Arvin et al. 2015 [12]	Active JPS: 40-90° Flexion	3D Motion-Capture System		1 and NS	RE: .75-.76 AE: .68-.80 VE: .20-.73
	Baert et al. 2018 [11]	Passive JPS: 20°, 45°, 70° Flexion	Analogue inclinometer	2 trial and two sessions within a day	2 and 3 hours training	Inter-rater reliability KOA: .88 without KOA: .82	Intrarater reliability KOA: .80-.93 without KOA: .76-.79
		KFS: 25%, 50%, 75% force	Handheld dynamometer			KOA: .53 without KOA: .08	KOA: .58-.78 without KOA: .00-.81
Ankle	Antcliff et al. 2021 [22]	AMEDA: Inversion	Platform	51 trial and 3 sessions	NS	Test-retest reliability was assessed with 13 old participants: .71 all years: .65
	Ko et al. 2015 [13]	TTDPM: Plantar & Dorsi Flexion	Isokinetic dynamometer	4 times a nd 2 days interval	NS	Test-retest reliability was assessed with 27 participants: .88
		Passive JPS: 5° Plantar and 5° Dorsi Flexion				Test-retest reliability was assessed with 27 participants: .44
	Westlake et al. 2007 [26]	TTVD: Plantar & Dorsi flexion	Isokinetic dynamometer	NS and 2 wks. interval	NS	Test-retest reliability was assessed with 8 old participants: .86

ICC = Intraclass Correlation Coefficient, JPS = Joint Position Sense, NS = Not stated, RE = Relative Error, AE = Absolute Error, VE = Variable Error, KOA = Knee Osteoarthritis, KFS = Knee Force Sense, AMEDA = Active Movement Extent Discrimination Apparatus, TTDPM = Threshold to Detection of Passive Movement, TTVD = Threshold to Velocity Discrimination

4. Summary of the measurement error data from the included studies

The measurement error was assessed in only two studies, using the standard error of measurement (SEM), minimal detectable change (MDC), and limits of agreement (LoA) (Table 5).

Table 5

Measurement Error data of the studies

Body Part	Study	Proprioception outcome	SEM		MDC		LoA
Hip	Arvin et al. 2015 [12]	JPS: Flexion	RE: .71°–1.03° AE: .63°–.72° VE: .66°–.94°		NS		RE: -2.99°, 2.74°(Lt.)/-2.05°, 1.9°(Rt.) AE: -1.86°, 2.16°(Lt.)/-1.43°, 2.06°(Rt.) VE: -1.83°, 1.81°(Lt.)/-3.01°, 2.17°(Rt.)
		JPS: Abduction	RE: .48°–.96° AE: .39°–.67° VE: .54°–.68°		NS		RE: -1.51°, 1.13°(Lt.)/-2.95°, 2.39°(Rt.) AE: -1.45°, .70°(Lt.)/-2.26°, 1.43°(Rt.) VE:-2.14°, 1.62°(Lt.)/-.89°, 2.09°(Rt.)
Knee		JPS: Flexion	RE: 2.20°–2.71° AE: 1.61°–1.68° VE: 1.42°–2.08°		NS	NS	RE: -6.82°, 5.36°((Lt.)/-7.12°, 7.89°(Rt.) AE: -4.49°, 4.44°(Lt.)/-4.06°, 5.26(Rt.) VE: -5.79°, 5.73°(Lt.)/-3.15°, 4.73°(Rt.)
	Baert et al. 2018 [11]	JPS: 20°, 45°, 70° Flexion	Inter-rater reliability KOA: 1 without KOA: 1	Intrarater reliability KOA: 1 without KOA: 1-2	Inter-rater reliability KOA: 3 without KOA: 3	Intrarater reliability KOA: 3-4 without KOA: 4	NS
		KFS: 25%, 50%, 75% force	Inter-rater reliability KOA: 7.4 without KOA: 10.7	Intrarater reliability KOA: 4.8-9.6 without KOA: 6.6-10.8	Inter-rater reliability KOA: 20.6 without KOA: 29.6	Intrarater reliability KOA: 13.2-26.7 without KOA: 18.3-29.9	NS

SEM = Standard Error of Measurement, MDC = Minimal Detectable Change, LoA = Limits of Agreement, JPS = Joint Position Sense, NS = Not stated, RE = Relative Error, AE = Absolute Error, VE = Variable Error, KOA = Knee Osteoarthritis KFS = Knee Force Sense

This systematic review aimed to identify reliable measurement tools for proprioception in the lower limbs. Nevertheless, a formal evidence synthesis or meta-analysis was not feasible because of the limited studies specifically focusing on lower limb proprioception in healthy adults aged 50 and older. The studies included also showed heterogeneity in methodologies, populations, and outcome measures, further complicating any attempt at quantitative synthesis.

1. Main Findings of the Review

This study analyzed proprioception measurement methods for lower limb joints by dividing them into hip, knee, and ankle joints, considering the differences in evaluation tools and methods reported for each joint. The purpose was to evaluate the impact of these variations on the study outcomes more accurately. Although various statistical indicators were reported during the reliability analysis, the ICC was the most consistently used reliability metric. According to Koo et al., ICC values > .90, .75 to .90, .50 to .75, and < .50 indicate excellent, good, moderate, and poor reliability, respectively [19]. The ICC was selected as the primary indicator for evaluating reliability because all the studies analyzed in this review used it.

1) Hip Joint Proprioception

Arvin et al. reported that the ICC for the relative error (RE) in hip abduction active JPS ranged from .81 to .93, while it ranged from .71 to .81 for the absolute error (AE), indicating moderate to excellent reliability, with a low SEM of less than 1° [12]. For hip flexion active JPS, the ICC for RE was .77 to .79, showing good reliability, but the AE had ICC values ranging from .11 to .56, indicating poor to moderate reliability. Arvin et al. attributed the low reliability of AE in hip flexion active JPS to the limited variance in the measured values, making it difficult to obtain a diverse range of data [12]. In addition, the AE values are calculated by converting negative values into absolute values, which may reduce variability, leading to lower reliability. Arvin et al. also reported that the cognitive load could affect proprioceptive testing, so they encouraged the participants to support their upper bodies to avoid distractions during the measurement [12].

2) Knee Joint Proprioception

Knee joint proprioception was evaluated using active JPS for knee flexion [12], passive JPS for knee flexion, and KFS [11]. Active JPS showed ICC values of .75 to .76 for RE and .68 to .80 for AE, indicating moderate to good reliability, whereas passive JPS showed ICC values of .76 to .93, indicating moderate to excellent reliability. In contrast, KFS had ICC values ranging from 0 to .81, showing a wide variation in reliability. The FS test, which assesses force perception and the participant's ability to reproduce a given force, may have shown lower reliability because of the influence of the evaluator's strength and weight when using handheld dynamometers, which are less reliable than the isokinetic dynamometers for assessing global muscle groups [20].

The knee passive JPS was assessed using the ICC(2,k) model, while active JPS was assessed using the ICC(2,1) model. Despite the differences in the models, the Passive JPS reported higher ICC values and lower SEM values, possibly because of the impact of body posture and cognitive load on proprioception, as considered by Arvin et al. [21,22]. The active JPS involved a physically challenging movement, which required older adults to stand, hold a bar with both hands, and bend one knee while raising the leg, potentially negatively influencing proprioceptive assessment [23]. In contrast, Passive JPS was conducted in the seated position, requiring less cognitive demand. Given the conflicting findings across studies regarding which method (active or passive) yields higher reliability, it is difficult to conclude that either method has a definitive advantage [24,25]. These results suggest that assessing proprioception in a standing position may be more appropriate if the goal is to simulate fall-related scenarios more closely, whereas seated assessments may be preferable when prioritizing reliability.

3) Ankle Joint Proprioception

Ankle joint proprioception was assessed using TTDPM, passive JPS [13], AMEDA [22], and TTVD [26]. Although JPS methods showed relatively high reliability for hip and knee joints, ankle passive JPS had an ICC of .44, indicating poor reliability [13]. The ankle is less sensitive to angular changes [22], and the small range of motion (5°) used in Ko et al.’s study may have made it difficult to detect changes, contributing to the low reliability despite the assessment being conducted in a seated position [13]. AMEDA had moderate reliability with an ICC of .71, while TTDPM and TTVD showed good reliability with ICC values of .88 and .86, respectively. Although Antcliff et al.'s simplified AMEDA method reduced the cognitive load by asking participants only to identify the direction of movement instead of memorizing five specific angles, it may have shown lower reliability than TTVD and TTDPM, which were measured in the seated positions and involved passive movement through isokinetic dynamometry, resulting in a lower cognitive load [22].

2. Clinical Practical Implications of Reliability of Lower Limb Proprioception Measurements

Although the studies reviewed used expensive equipment, such as isokinetic dynamometers, 3D motion-capture systems, and platforms to measure proprioception accurately [22,12,13,26], Baert et al.'s study is noteworthy for using an affordable analogue inclinometer [11]. Cost-effective and straightforward proprioception measurement methods are essential when older adults live in group settings or community environments. The inclinometer used in Baert et al.’s study also maintained high reliability, making it suitable for proprioception measurements in older adults [11]. The evaluators were given three hours of training to measure the passive JPS using the inclinometer. The results showed high reliability with low SEM values. These findings suggest that the inclinometer may be one of the most appropriate tools for use in clinical and community settings, offering high reliability at a low cost.

The systematic review by Ager et al., who evaluated the reliability of various proprioception measurement tools and ranked them by the ICC, supports the notion that the inclinometer is a cost-effective yet reliable option among low-cost tools. The highest to lowest-ranked tools were the isokinetic dynamometer, continuous passive motion (CPM) device, fabricated lab systems, AMEDA, inclinometer, photo analysis, laser pointer, iPod touch, and goniometer [14]. Although the inclinometer ranked in the middle, it showed higher reliability among low-cost tools. This study focused on the shoulder joint; hence, a similar trend could be expected in lower limb proprioception measurements. Therefore, the inclinometer provides a cost-effective yet reliable option, making it a practical choice for clinical and community settings.

3. Recommendations for Future Research and Limitations of the Review

This review highlights the need for further research to clarify the effects of cognitive load and postural changes on proprioception measurements in older adults. Furthermore, there remains a lack of high-quality studies on reliable proprioception measurement methods for older adults, and more active research in this area is warranted. Developing reliable assessment tools would provide essential evidence for the early detection and management of fall risks in older adults in clinical and community settings.

One limitation of this study was the small number of studies included, which may restrict the generalizability of the results. Moreover, one study included in the review received a "worst score counts" rating on the COSMIN risk of bias scale, and none of the studies received an overall "very good" rating in all categories, indicating potential quality limitations. Lastly, the small sample sizes used in the reliability studies on older adults may also limit the generalization of the findings.

This systematic review comprehensively analyzed studies on proprioception measurement methods for lower limb joints. By evaluating the hip, knee, and ankle joints separately, differences in reliability were found depending on the joint, with various assessment tools and methods used across studies. In particular, while expensive equipment like isokinetic dynamometers provide high reliability, there is a growing need for cost-effective yet reliable tools for use in clinical and community settings. In this regard, tools such as the analogue inclinometer present a practical alternative, showing that sufficient reliability can still be achieved.

No financial support was received to conduct this systematic review.

The authors declare no conflicts of interest related to this systematic review.