Assessing Proprioception

Abstract

Proprioception is a vital aspect of motor control and when degraded or lost can have a profound impact on function in diverse clinical populations. This systematic review aimed to identify clinically related tools to measure proprioceptive acuity, to classify the construct(s) underpinning the tools, and to report on the clinimetric properties of the tools. We searched key databases with the pertinent search terms, and from an initial list of 935 articles, we identified 57 of relevance. These articles described 32 different tools or methods to quantify proprioception. There was wide variation in methods, the joints able to be tested, and the populations sampled. The predominant construct was active or passive joint position detection, followed by passive motion detection and motion direction discrimination. The clinimetric properties were mostly poorly evaluated or reported. The Rivermead Assessment of Somatosensory Perception was generally considered to be a valid and reliable tool but with low precision; other tools with higher precision are potentially not clinically feasible. Clinicians and clinical researchers can use the summary tables to make more informed decisions about which tool to use to match their predominant requirements. Further discussion and research is needed to produce measures of proprioception that have improved validity and utility.

Introduction

What Is Proprioception: Ambiguity Leading to Confusion

In the early 1830s, Sir Charles Bell described the sixth sense, referring to the sense of position and action of the limbs. ¹ Proprioception was further discussed by Sherrington in his seminal text and lecture series some 70 years later. ² Since then the term has been used to describe a variety of senses and therefore has become somewhat ambiguous. The word proprioception comes from the Latin proprius, meaning “one’s own,” combined with the concept of perception: thus a literal translation is that of “perceiving one’s own self.” This notion of self-perception reflects one’s ability to have both a sense of body orientation and position as well as a sense of body and limb motion. Accordingly, words often used interchangeably with proprioception are (joint) position sense, kinesthesia, movement sense, body position in space, sense of effort, or sense of force. The confusing and inconsistent application of these terms, particularly in the clinical domain, reflects the slowly emerging understanding of the nature of how we sense ourselves.

Berthoz ³ classically defines kinesthesia as the sense of movement, of which proprioception (sense of position and of velocity) is a part, along with cutaneous receptors, and the vestibular and visual systems. Conversely, other authors cite proprioception as having 3 submodalities: kinesthesia, joint position sense, and sensation of force (see, eg, Niessen et al ⁴ ). Others again add body segment static position, displacement, velocity, acceleration, and muscular sense of force, effort, or heaviness (see Ogard ⁵ or Proske and Gandevia ⁶ ) to the list of proprioceptive constructs. In this review, we will refer to proprioception as a collective term for these “subsenses,” unless the evidence only pertains to one in which case we will refer to that subsense individually.

Finally, proprioception can be considered one of the subsystems within the somatosensory system (along with pain, touch, and thermal sensation) and has also been considered interoceptive in that the sensory information is derived from changes within internal structures. This classification is in distinction to exteroception where the stimulus originates from outside the body such as external heat for thermoreception or light stimuli for vision.

Proprioceptive Receptors

Proprioception, or proprioceptive acuity, is a complex system that involves both peripheral and central systems. The evidence for the prime proprioceptive receptor favors muscle afferent input, ⁶ in particular muscle spindles. These receptors are specialized fibers within the muscle that detect change in muscle length and also the velocity of contraction ⁷ (or body part motion as the first derivative of length, ie, the rate of the change in length). If a passive lengthening is applied to a muscle, spindle exafferent signals are produced and interpreted as a sensation of movement, with increasing velocities causing an increasing response. ⁶ The spindle system has the capacity to anticipate length change because it can detect velocity as well as length (which changes more quickly). ³ Furthermore, motion direction can be perceived relative to which particular muscle has shortening or lengthening activity, and presumably by comparison between agonist and antagonist activity ratios, joint position can be perceived. The muscle spindle is also under fusimotor control (the gamma system), during contractions, which has the capacity to alter the calibration or sensitivity of the receptor by altering its internal length. ⁸ This modulation allows adaptation of the receptors during action and also allows simulation in the absence of real action. ³ Related to the sensitivity of the spindle is the role of thixotrophy—the phenomenon that demonstrates a relationship between a muscle’s properties and its immediate history of contracting/lengthening. It is a property of muscle (and muscle spindles) and influences proprioception—that is to say proprioception (spindle sensitivity) can be altered, dependent on whether the muscle has recently contracted or not (see Proske and Gandevia ⁶ )

Proske and Gandevia ⁶ summarize the evidence that receptors in the skin (cutaneous receptors) also contribute to joint position and motion sense, for example, as skin strain, particularly at the digits, elbow, and knee. Receptors analogous to the cutaneous receptors also exist in joint structures. For example, the more superficial Ruffini endings in the joint capsule, ligaments, and menisci are slow adapting mechanoreceptors. This allows detection of static joint position, intra-articular pressure, and possibly joint motion in terms of amplitude and velocity. ⁸ Pacinian corpuscles are deeper in these joint connective tissues with a lower mechanical threshold and are faster adapting—allowing them to be more responsive to changes in velocity, that is, acceleration and deceleration. Lephart and Fu ⁸ also describe free nerve endings widely distributed in articular structures and which may play a role in detecting severe mechanical deformation or inflammatory changes. However, it is now accepted that these mostly mechanical receptors predominate at extreme angles of joint position and motion and are relatively silent midrange.^6,9 Golgi tendon organ-like receptors are also found in cruciate and collateral ligaments and menisci; like the mechanoreceptors listed, it is reported that they are useful at extremes of range as limit detectors. ⁸ This then leaves the muscle spindles to provide most information in the middle region of joint action. ⁶

Effort-based signals, that is, the efferent commands to muscles, have a role in proprioception (particularly sense of force and/or position sense). These signals are distinct from the traditional kinesthetic senses in that they are based on motor commands and therefore not experienced in a passive limb. However, Gandevia et al ¹⁰ investigated paralyzed and anesthetized upper limbs to demonstrate an alteration in position sense at the wrist joint, relative to the effort of attempting to produce a contraction in the paralyzed muscle(s), concluding there is a definitive role for “outflow” (efferent) signals in proprioception.

In summary, proprioception is based on an ensemble of sensory inputs that serve sensing, producing, predicting, and simulating joint position, joint motion (velocity and direction), and force specification. ³ There is considerable specificity and sensitivity in this ensemble arrangement as well as redundancy, particularly when proprioception is converged with the visual and vestibular systems in detecting motion and spatial orientation (from visual- and gravity-referenced coordinates).^3,11 The most basic of postural control tasks, such as standing in regular environments, ¹² is contingent on the coordination of these processes.

Clinical Implications of Proprioception

In motor control, proprioception, along with the other senses, is important in both feedback and feedforward operations and can be used in combination or in the absence of other sensory systems. Proprioceptors have a role in motor planning (feedforward for anticipation, preparation, and response planning) as well as rapid wiring into adaptation mechanisms to effect performance changes during task execution (feedback). Clinical scenarios where proprioception is lost or degraded classically result in loss of movement control where the person must then rely on visual input for feedforward and feedback processes. This may result in difficulty in learning novel movement, and also difficulty in improving the quality of movement or maintaining quality over a series of repetitions because of the absence of feedback for adaptation and skill refinement. ¹² Not only are dexterous tasks affected but also balance and locomotion despite the high degree of redundancy with vision and vestibular input for these activities. Age has an impact on proprioception ¹³ ; proprioception lags somewhat behind vision in an infant’s development and then declines with age, particularly beyond 50 years.

Clinically, reduced proprioception has been most obviously implicated in stroke, ¹⁴ age-related falls, ¹³ and peripheral neuropathy ¹⁵ and also in movement disorders such as Parkinson’s disease, Huntington’s disease, and focal dystonia. ¹⁶ There has also been a lot of interest in the role of suboptimal proprioception in the etiology and/or rehabilitation of orthopaedic ¹⁷ and sporting injuries. ¹⁸ Poor proprioception has been reported in pain states such as chronic low back ¹⁹ and in persistent whiplash and associated disorders, ²⁰ and in other diverse conditions such as developmental coordination disorder, ²¹ hypermobility syndrome ²² and Asperger syndrome. ²³

Measuring Proprioception

There is clearly no single measure of proprioception due to the complexity of the neurophysiological processes that encompass proprioception described above.

In order to investigate the impact of a loss of proprioception on function, or the degree of loss from pathology, clinicians and researchers have developed many ways to capture proprioceptive acuity. More recently this has been driven by inquiry into the capacity of the proprioceptive senses to be improved through sensory-specific training or upregulation via afferent stimulation. ²⁴

Tests have differentiated between the 2 main proprioceptive functions—detection of static position and detection of motion. The latter is further broken down into the threshold of motion detection, that is, the threshold amount/speed of motion required for detection to occur, and second, the direction of motion (eg, flexion or extension), which is considered a discrimination task. These 2 tests are usually performed passively and clinically and are administered to the great toe in a nonstandardized fashion. Detection of position has been performed by position copying or position matching tasks that can be done actively or passively. In order to further refine testing, attempts are often also made to reduce cutaneous stimulation during motion or position detection. However, given the multiplicity of constructs attributed to proprioception and the array of physiological processes, such tests often raise more queries than answers.

We embarked on this review systematically as the first stage in clarifying the ambiguities discussed above. We chose a systematic approach to cover the greatest breadth and depth of the literature. As such the aims of this systematic review were to

Identify the reported methods of assessing proprioception in healthy and pathological populations

Classify the subsenses of proprioception they purport to measure (external validity)

Report the clinimetric properties investigated for each tool (internal validity)

The overall objective was to provide clinicians and clinical researchers with a summary document of tools to enhance selection relative to need, with an explanation of the limitations or issues in the identified current approaches.

Methods

To identify proprioception assessment tools, a systematic search of the literature was conducted. The search strategy involved 2 steps. First, a comprehensive electronic database search was conducted. Then a secondary search was conducted looking at the reference lists from articles that were reviewed in full-text from the database search.

The databases searched included EBSCO, Journals@OVID, AMED, EMBASE, Medline, Ovid Medline, Science Direct, SportDiscus, E-journal, Ageline, CINAHL, and Highwire Press. Dates were from database inception and the searches were completed in December 2013. Search terms used included combinations of proprioception, kinesthesia or joint position, joint motion, with terms such as tests, clinical measures, clinical examinations, diagnostic accuracy, predictive value, sensitivity or specificity.

Articles identified in the database search were then evaluated by a team of researchers to ensure the research article and tools of proprioception assessment met the following inclusion criteria:

Explicitly described and employed a tool or tools to measure proprioception (or allied terms)

In any human population

Written in English

Two researchers (SH and a research assistant) agreed on eligibility for inclusion. There was no attempt to critically appraise the quality of the individual articles as they were of various study designs not necessarily related to the use of the tool itself. For example, the primary article may have been a randomized controlled trial using the tool as an outcome measure or it may have been a single case study using the same measure. We determined the level of study design itself did not offer any indication of the robustness or utility of the tool itself and therefore was not relevant.

The articles were then collated into similar tools with similar targeted subsenses (relative to aim 2) and data extracted to describe the joints measured, the construct, and the broad method of testing. Further data regarding population tested, equipment, and procedures were also extracted. The subsense or construct clusters were identified as

Joint position detection—active/instantaneous position or passive (AJPD or PJPD) (acknowledging the position is enacted AT the joint, not necessarily detected BY the joint)

Passive motion detection threshold (PMDT)

Passive motion direction discrimination (PMDD)

To fulfil aim 3, further data were extracted from the articles retrieved (and from the reference lists) to record relevant clinimetric properties, based on the process used by Slater et al. ²⁵ See Appendix 1 for an example of the data extraction pro forma. The properties scrutinized were the following:

Reliability: test–retest, reproducibility, intra- and interrater and internal consistency

Validity: face, criterion, content, construct, concurrent, or factor analysis

Responsiveness: sensitivity, specificity, floor and ceiling effects, discrimination

Precision: accuracy, standard error

To further inform potential readers of the clinical features of the tool (also aim 3), we also considered subjectively

Client-centered attributes: appropriate, meaningful, acceptable

Tester-centered attributes: feasibility and clinical utility—ease of use, clinically meaningful, ease of interpretation, normative scores

Two researchers completed this data extraction phase and crosschecks occurred on a random audit basis.

Results

We will first report the results of the search and then results relative to the 3 aims of the study.

Step 1, which involved the database search, ultimately yielded 40 articles from the initial database search list of 935, with the full process of inclusions and exclusions detailed in Figure 1. Step 2, which involved pearling the reference lists of the articles retrieved for full-text review, revealed a further 17 articles to be retained for final inclusion, giving a total of 57 articles.

Figure 1.

PRSIMA flow chart of article selection.

Identification of Tests

Table 1 summarizes the 57 research articles included in this review. From these 32 tests of proprioception were identified (that appeared to be sufficiently distinct in relation to purported constructs, administration, and/or joints) and data extracted as detailed in the methods.

Table 1.

Summary of Identified Proprioception Tests.

Tool and Author/Date	Body Part, Construct, and Broad Method	Population	Equipment	Procedure
Lumbar proprioception equipment (LPE) Zazulak et al26 (2007) and previously described by Taimela et al 27 (1999)	Trunk AJPD PJPD Active repositioning and passive recognition (joint angle error)	Collegiate athletes 26 Population with lumbar fatigue 27	Motorized apparatus fixes upper body and rotates lower body around vertical pivot through L4/5	Active and passive repositioning—apparatus rotates lower trunk and then subject indicates regaining of original neutral position either actively OR when passive motion reaches neutral position (lying)
Spinal Motion Apparatus (SMA) Pankhurst and Burnett 17 (1994) (also designer)	Lower back PJPD PMDT PMDD Passive recognition (joint angle error) Threshold for detection Direction accuracy	Male firefighters with and without back injuries	Motorized apparatus that produces passive motion of Lx in 3 planes (upper body fixed)	Apparatus moves lower body passively in any 1 of the 3 planes—person detects motion; picks direction of motion; has to identify test place from neutral (lying)
Active movement extent discrimination apparatus (AMEDA) Hobbs et al 28 (2010)	Lumbar AJPD Discriminate position differences in 11-19° Lx flex	Healthy individuals; with spinal pain and radiographic evidence of disc degeneration; post disc replacement surgery	Unrestrained standing (WB) with stopper at 5 preset (demonstrated) distances (stepper motor). Version for ankle with variable plate	Person had to distinguish between 5 preset active flexion positions as set by concealed stopper (standing)
Neck proprioception testing device (NPTD) Lee et al 29 (2005)	Cervical AJPD Discriminate position differences in Cx rotation (25-41°), Cx retraction (1 to 1.8 cm)	Healthy students (subclinical neck pain)	Cervical version of AMEDA above. Unrestrained cervical motion with stopper at 5 preset positions (stepper motor)	Person had to distinguish between 5 preset active rotation and retraction positions as set by concealed stopper (sitting)
Proprioceptive apparatus (no name) Cuomo et al 30 (2005)	Shoulder PJPD PMDT Passive recognition (joint angle error) Threshold for motion detection	Patients with unilateral advanced glenohumeral arthritis before and after total shoulder arthroplasty; healthy controls	Motorized apparatus to produce motion at the GH in 3 planes	Subjects arm was passively moved by device through range and subject indicated when predetermined position attained; second test subject indicated when onset of passive motion was detected (seated and standing)
“Manipulandum” Bevan et al 31 (1994) (previously described by Cordo et al 32 1994)	Elbow PJPD Passive recognition (joint angle error) Passive recognition (distance estimation error)	Health subjects	Elbow (forearm and upper arm) strapped to motor device that produced passive change of joint angle (verified by goniometry)	Person indicates when their elbow reaches one of the pretrained joint angles (flex/ext); indicates when their forearm has moved the pretrained specified distance (sitting)
“Kinarm” Bhanpuri et al 33 (2012)	Elbow PJDT PMDT Passive detection of motion then discrimination of magnitude	Cerebellar subjects	Elbow (forearm and upper arm) strapped to exoskeleton robot system that produced passive change of joint angle	Subjects report if they feel a movement or not (by robot). Second movement elicited and subject indicates if greater or less than first
“Manipulandum” Wong and Henriques 34 (2009)	Upper limb PMDD Direction accuracy	Healthy subjects	Robot manipulandum that passively moves hand through different pathways (vision obscured)	Distinguish between 2 tilting pathways and 2 curved pathways (sitting)
“Manipulandum” Simo et al 35 (2011)	Upper limb PMDT Passive motion detection	Healthy and stroke impaired	Robot manipulandum that produces arm displacements (vision obscured)	Distinguish movement in 1 of 2 time periods (other stationary)
Shuttle Miniclinic constant-resistance device Lin et al 36 (2006)	Hip and knee PJPD Passive recognition (joint angle error)	Normal males—elite, amateur, and novice tennis players	Constant resistance device applied to sole of foot to allow increase/decrease hip and knee joint angles in closed chain electro-inclinometer measured thigh and shin motion	Subject presses on device to extend limb from starting position of 60° hip flex/90° knee flex to a pretrained target position and repeat (supine)
Proprioceptive apparatus (no name) Friden et al 37 (1996); Fischer-Rasmussen and Jensen 38 (2000); Barrack et al 39 (1983); Corrigan et al 40 (1992); Co et al 41 (1993) Also Barrack et al 42 (1989) = deviser	Knee AJPD PMDT Active reproduction (joint angle error) and also reproduces visually with object Threshold for motion detection	Healthy subjects; subjects with ACL deficient knees ACL reconstructions; bilateral OA knees; elderly non-OA 43 Health female gymnasts 44 OA knees 45	Motorized apparatus to passively move shin to change knee joint angle; in some versions, motor can be disengaged to allow active movement	Subject detects onset of passive motion/change of position at knee; actively reproduces target demonstrated angle; indicates target angle with goniometer (side-lying using underneath leg or supine); actively matches contralateral leg to match angle of test knee; indicates onset of (slow) passive motion (sitting)
Proprioceptive device (no name) Brindle et al 46 (2010)	Knee/ankle PMDT PMDD Passive detection of motion onset and direction	Healthy subjects	Motorized apparatus (Biodex) that passively moves both knee and ankle	Subject’s foot rotated to lengthen/shorten gastrocnemius, while knee flexed/extended. Indicate onset of movement and direction
Movement detection apparatus Matre and Knardahl 47 (2003), further described by Matre et al 48 (2002)	Ankle PMDT PMDD Threshold for motion detection Direction accuracy	Healthy subjects	Motorized rotating platform, axis aligned with ankle joint	Subjects foot rotated in dorsi- or plantarflexion direction—indicate when motion was detected AND which direction (seated)
Passive ankle joint repositioning test Fu et al 18 (2007)	Ankle PJPD Passive recognition (joint angle error)	Health male basketball players	Cybex motorized apparatus to produce passive ankle displacement at constant speed; electrogoniometry	Subjects foot rotated in plantarflexion direction—indicated when predetermined target was reached (prone)
Cervicocephalic kinesthesia Kristjansson et al 49 (2001) tests from multiple authors Also Cervical Joint position error (JPE) Treleaven et al 50 (2003); Treleaven et al 20 (2006) Also JPE-Torsion Chen and Treleaven 51 (2013)	Cervical AJPD Active reproduction (joint angle error)	Healthy volunteers 49 Healthy controls and subjects with WADS (whiplash)50,51	Polhemus 3space Fast track (sensors)—calculated error, or laser marker 51	Test 1: relocation of head to natural head posture (NHP) (after active rotation left and right) Test 2: Active relocation to 30° rotation from NHP (after passive criterion) Test 3: Preset passive trunk rotation 30°, subject repositions head to NHP and then back to original 30 relative Test 4: Figure of eight motion and relocation to NHP (active) Test 5: Figure of 8 movement—accuracy of passing through NHP at each crossing of 8 JPE-Torsion—modification of rotating trunk “under” fixed head to differentiate from vestibular stimulation
Thoracolumbar proprioception Koumantakis et al 52 (2002) (original protocol by Gill and Callaghan, 19 1998)	Spinal AJPD Active reproduction (joint angle error)	Individuals with and without low back pain	Lumbar motion monitor (exoskeleton) to measure absolute error (between pelvis and chest harnesses)	Active target reproduction in 3 planes after initial active criterion position (flex/rotation/lat flex)
Arm position-matching task Dukelow et al 53 (2010); Debert et al 54 (2012)	Upper extremity AJPD Active reproduction (spatial coordinates)	Healthy versus people with stroke 53 or traumatic brain injury 54	KINARM exoskeleton on both arms—to apply load/measure motion	Passive positioning of arm, actively matched by contralateral arm
Active Knee joint reposition sense Bullock-Saxton et al 13 (2001)	Knee AJPD Active reproduction (joint angle error)	Young, middle age, and old healthy subjects	Polhemus 3space Fast trak (sensors)	Sensors detect position of test limb in active criterion versus subsequent active reposition under FWB and PWB conditions, and error is calculated
Active Knee joint reposition sense Kramer et al 55 (1997)	Knee AJPD Active reproduction (joint angle error)	Healthy and with patellofemoral pain syndrome	Electro-goniometer to measure absolute error for reposition	Active knee repositioning to target criterion angle in sitting (NWB) and standing (squatting—WB)
Active knee joint reposition test Marks et al 56 (1993)	Knee AJPD Active reproduction (joint angle error)	Women with healthy versus OA knees	Photographic capture of absolute error in repositioning from criterion (from limb markers)	WB—stood on test limb and tried to actively reproduce criterion angles NWB—stood on contralateral leg and bent test leg (open chain) to criterion angle
Limb copying and reproducing tests Kaplan et al 57 (1985)	Knee AJPD Active reproduction (joint angle error) ipsi- and contralateral	Healthy older and younger women	Goniometer to measure error between reproduction and criterion	1. Knee passively positioned and contralateral actively reproduces 2. Reproduce criterion position in ipsilateral knee at 15 seconds and 60 seconds after demonstration
Limb copying test As reported in various including Paqueron et al 58 (2004); Goble et al 59 (2012)	Elbow AJPD Active reproduction (joint angle error) contralateral or ipsilateral	Anesthetized UL 58 Adults with cerebral palsy 59	Observation	Limb moved passively and while shielded from vision. Opposite limb “copies” actively OR ipsilateral limb copies after period of time (requires spatial working memory). Improves with longer encoding time for example position.
Joint position sense Reported variously in Lord et al60-62 (1991, 2000, 2002); Sturnieks et al 63 (2004); Whitney et al 64 (2005); Tiedmann et al 65 (2007); Wood et al 66 (2008) Apparatus design by De Domenico and McCloskey 67 (1987)	Lower limb AJPD Active reproduction (joint angle error) contralateral	Healthy aged; Parkinson’s; older with arthritis	Vertical Perspex sheet—error of matching measured by inscribed protractor in degrees	Limb on one side placed on sheet (eg, big toe) and simultaneously other leg positioned to touch corresponding spot on sheet from other side (eyes closed)
Proprioceptive assessment: Modified Thomas splint with Pearson knee flexion piece Reported variously in Warren et al 68 (1993); Barrett et al 69 (1991)	Knee PJPD Reproduction (joint angle error) on mock-up leg	Adults anterior cruciate reconstruction; 18 normal versus osteoarthritic and replaced knees; 69 knee OA/arthroplasty 68	Thomas splint with a Pearson knee-flexion piece was modified to provide well-padded support to whole leg. Protractor attached to knee piece. Axes aligned. Warner added electronic goniometry and smoother positioning (limits sensation to capsular and ligamentous joint receptors).	Supine and leg screened from view. Leg moved passively to 1 of 10 predetermined positions of flexion. Subject manually reproduces angle on a mock-up leg with goniometry. Error between real knee angle and perceived angle is measurement
Rivermead Assessment of Somatosensory Perception (RASP) Winward et al 70 (2002); Tyson et al 14 (2008)	Elbow, wrist, thumb, ankle and big toe proprioceptive subtests PMDT PMDD Threshold for motion detection Direction accuracy	Adults with stroke	Nil	Participants eyes closed, joint moved passively into flexion or extension: for detection indicate when movement occurs; for discrimination indicate which direction. Score system 0-1 absent; 2-5 impaired; 6 intact
Cumulative Somatosensory Impairment Index (CSII) Deshpande et al 71 (2010)	Lower extremity Proprioceptive subtest AJPD Reproduction (joint angle)	Healthy controls, diabetic, peripheral arterial disease, stroke	Nil	Reference ankle positioned by examiner and subject matches position (10/20 degrees or neutral)
Distal Proprioception Test (DPT) Richardson 15 (2002)	Toe, ankle PMDT PMDD Threshold for motion detection Direction accuracy Note these tests can also be modified to simply detect position up/down AFTER passive motion, ie, PJPD	Peripheral neuropathy, control	Nil	Grasp lateral borders of great toe and perform up/down movements first with eyes open then eyes closed—test is 10 small amplitude random, smooth movements over a distance of app 1 cm and speed of 1 p/s. Score for correct perception of movement and identify direction. <8/10 trials positive.
Dual Joint Position Test (DJPT) Beckman et al 72 (2013)	Digits PJPD Direction accuracy (simultaneous in 2 digits)	Healthy controls and those with lemniscal system dysfunction (MS or vasculitis)	Nil	As for DPT but move 2 digits simultaneously—combination both up, both down, or one up/one down
Standardized sensory assessment for children Cooper et al 73 (1993)	Finger PMDT PMDD Threshold for motion detection Direction accuracy	Typically developing children (school-aged)	Nil	As for DPT but with motion occurring at the MCP of the finger, scored x/5 to discern direction of motion up/down
Thumb finding test Smith et al 74 (1983)	Thumb/upper limb PJPD Accuracy in locating passively placed limb (thumb)	Adults with stroke	Nil	Not stated. Assumed to be passive positioning of the affected thumb in different spatial coordinates and person has to make contact with test thumb using other (unaffected) hand. Score x/10
Wrist Position Sense Test Carey et al 75 (1996)	Wrist PJPD Accuracy in indicating joint angle	Adults with stroke	Splint device for forearm/hand within obscuring box; overlaid with protractor	Tester positions subject’s wrist in predetermined positions of flexion/extension; subject indicates best matched joint angle using protractor arms

Abbreviations: AJPD, active joint position detection; PJPD, passive joint position detection; PMDT, passive motion detection threshold; PMDD, passive motion direction discrimination; L or Lx, lumbar; WB, weight bearing; Cx, cervical; GH, gleno-humeral; flex, flexion; ext, extension; ACL, anterior cruciate ligament; OA, osteoarthritis; WADS, whiplash and associated disorders; NWB, non–weight bearing; MS, multiple sclerosis; MCP, metacarpophalangeal joint.

The body part most commonly measured was the knee (11) followed by 4 tests identified for the lower trunk or back, 4 for the ankle, and 2 distinct protocols for the cervical spine. Other areas were often tested as a composite (eg, hip and knee and ankle in weight-bearing), or targeted distal joints of upper and lower limbs (eg, toes and fingers/thumb).

The tests were used on widely varied populations. The majority were with healthy adults or used healthy adults as a control. Three looked at populations of athletes (gymnasts and tennis and basketball players), a specific profession (firefighters with and without lower back injuries), or the influence of age (children, adults, and elderly). The majority of pathological populations had musculoskeletal issues such as anterior cruciate ligament deficiency (ACL; 4), osteoarthritis (5), or pain (4). Postsurgical comparisons were also featured, for example, pre- and post-ACL reconstruction (3), shoulder arthroplasty, or knee replacement. Neurological pathologies were less well featured with 4 tests described in stroke and 1 each for Parkinson’s disease, traumatic brain injury, cerebral palsy, peripheral neuropathy, and anesthetized limbs.

Proprioceptive Subsenses Tested

Table 2 summarizes the tests in relation to what they test (the subsenses or constructs). The majority of tests assessed joint position detection either actively or passively (AJPD or PJPD) through recognition or reproduction tasks with the measurement being joint angle error (JAE; ie, the difference between the true target angle and the reproduced angle or angle at which recognition occurred). There was variation in how the target angles were demonstrated (on the limb itself, or the contralateral limb, using an angled lever) and whether the JAE was measured via motion capture equipment, electro-inclinometry, photographic capture, Perspex protractor grading, or goniometry (electric or manual). Various means were used to produce the passive motion for PJPD from motorized apparatus or robotics, to modified splints moved by the tester to simply the tester moving the body part. Where the task was active reproduction of the target angle, it was mostly reproduced on the ipsilateral side, but in the more manual-based tests it was reproduced on the contralateral side or on a sheet of paper, a goniometer, or a simulated limb.

Table 2.

Proprioceptive Tools Clustered Within 4 Identified Subsenses.

Proprioception Subsense	Physiological Basis 6	Tool(s)	Strengths	Limitations	Populations	Body Part Measured
Active joint position detection (subject actively reproduces established position)	Muscle spindle (major source) both afferent and efferent activity Skin stretch receptors (Ruffini) Joint receptors (limit detectors)	Lumbar proprioception equipment26,27 Active movement extent discrimination apparatus 28 Neck proprioception testing device 29 Proprioception apparatus37-45 Cervico-cephalic kinesthesia/JPE20,49-51 Thoracolumbar proprioception19,52 Arm positioning-matching task53,54 Active knee joint reposition sense14,55 Limb copying and reproducing tests57-59 Joint position sense60-67 Cumulative somatosensory impairment index 71	Reflects functional use of proprioception—gauging position in space through volitional movement No extraneous sensory input from passive device (pressure sense)	More likely to be limited by pain Requires kinesthetic memory of preestablished position Requires sufficient motor control	Collegiate athletes Lumbar fatigue Spinal pain/surgery Subclinical neck pain ACL deficiency/reconstruction OA Elderly Whiplash Low back pain Stroke/TBIPatella-femoral pain Cerebral palsy Anesthetized Parkinson’sDiabetic Peripheral arterial disease	Lumbar spine Cervical spine Knee Upper limb Elbow Lower limb Ankle
Passive joint position detection (subject recognizes established position when reproduced passively)	Muscle spindle (major): mean rate of background discharge generated by both primary and secondary endings Skin stretch receptors (Ruffini) Joint receptors (limit detectors)	Lumbar proprioception equipment26,27 Spinal motion apparatus 17 Proprioceptive apparatus 30 “Manipulandum”31,32 Kinarm 33 Shuttle Miniclinic constant-resistance device 36 Passive ankle repositioning test 17 Proprioceptive assessment68,69 Dual joint position test 72 Thumb finding test 74 Wrist position sense test 75	Does not require active control	Possible additive sensory input from passive device (pressure sense or auditory stimuli) Requires kinesthetic memory of preestablished position	Collegiate athletes Lumbar fatigue Male firefighters +/− back injuries Glenohumeral arthritis Cerebellar Elite/amateur/novice tennis players Basketball players Knee pathologies Lemniscal system dysfunction Stroke	Lumbar spine Shoulder Elbow Hip Ankle Knee Digits Wrist
Passive motion detection threshold (the threshold at which the subject can detect a moving state from the stationary)	Muscle spindle (major): primary endings stimulated by change in length and rate of change in length Skin mechano receptors (Meissner, Pacinian, Merkel and Ruffini endings) Joint receptors (limit detectors)	Spinal motion apparatus 17 Proprioceptive apparatus 30 Kinarm 33 “Manipulandum” 35 Proprioception apparatus37-45 Proprioceptive device 46 Movement detection apparatus47,48 Rivermead Assessment of Somatosensory Perception 70 Distal proprioception test 15	Does not require working memory	Detection thresholds higher for slower movements	Male firefighters +/− back injuries Glenohumeral arthritis Cerebellar Stroke ACL deficiency/reconstruction OA Elderly Stroke Peripheral neuropathy	Lumbar spine Shoulder Upper limb/hand and knee Knee and ankle Ankle Thumb Toe 1
Passive motion direction discrimination (subject discriminates between different movement directions—one or more axes of rotation)	As above	Spinal motion apparatus “Manipulandum” 34 Proprioceptive device 46 Movement detection apparatus47,48 Rivermead Assessment of Somatosensory Perception 70 Distal proprioception test 14 Standardized sensory assessment for children 73	Higher order test to not only detect motion but to discriminate direction	Response can be dichotomous and therefore easy to “guess,” eg, flexion versus extension	Male firefighters +/− back injuries Stroke Peripheral neuropathy Children	Lumbar spine Hand Knee and ankle Ankle Thumb Toe 1 Elbow Wrist Finger

Ten tests^{15,17,30,33,35,37,45-48,70} measured the threshold for detecting motion (PJMD) again using a variety of ways to produce the passive motion as described above. The most common measure was the error between the angle of actual commencement of motion and the detection angle. However, the more simple clinical tests simply required the person to detect motion compared to static position and were scored dichotomously for 5 to 10 trials.

Six tests^{15,34,46,48,70,73} also incorporated the direction of motion discrimination task (PMDD) by requiring the person to indicate whether the limb moved in a positive or negative direction relative to the defined plane of motion. Generally, the planes of motion were restricted to one but a few tests did incorporate up to 3 planes and one used “pathways,” that is, tilting or curved.

In all examples found, some attempt was made to reduce the influence of confounding senses though the degree to which this was attempted also varied, for example, all tests excluded vision while the most elaborate tests produced passive motion with pneumatic splints so that there was no discernment of pressure to imply motion or motion direction (via cutaneous receptors) or even used background white noise to cancel out any clues to motion commencement from the sound of the motor (see Table 2 for limitations).

Clinimetric Properties

Generally, the clinimetric properties for the identified tools were poorly evaluated or reported. Table 3 summarizes each test and the reported properties we were able to find, as well as our judgments on the user- and client-centered aspects. Six tests^{29,30,31,32,34,46,74,75} had no recorded properties tested. For the others, the main data available related to the discriminant ability of the test, that is, to differentiate between 2 groups (one normal and one pathological for example). The Rivermead Assessment of Somatosensory Perception (RASP) ⁷⁰ had the most reported properties and achieved “good” or higher in the evaluations. Furthermore, this test can be applied with no equipment to 5 different joints for greater clinical utility; however, it was not considered precise in that the scoring for each test move is only dichotomous and there is the confounder of the tester manually moving the body part with cutaneous input. The Modified Thomas splint with Pearson knee flexion piece^68,69 has been used and tested more widely with high values for test–retest and concurrent validity and ability to discriminate between different pathological groups and age ranges. The equipment requirements are not onerous but this device only assesses the knee.

Table 3.

Clinimetrics of Proprioception Tests From Published Literature.

Tool With Author/Date	Summary	Reported Psychometrics	Client-Centered Attributes	Tester-Centered Attributes
Lumbar proprioception equipment (LPE) Zazulak et al 26 (2007) and previously described by Taimela et al 27 (1999)	Trunk AJPD PJPD	Reproducibility of both active and passive: ICCs 0.61 and 0.58	Acceptable for research population Rotation motion is semifunctional	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Spinal Motion Apparatus (SMA) Pankhurst and Burnett 17 (1994) (also designer)	Lower back PJPD PMDT PMDD	Accuracy of motion within 2 mm range (within 0.1°) Reliability of PMDT ICC 0.46 to 0.92 Reliability of PJPD ANOVA no sig. differences	Acceptable for research population Three planes to cover more functional motion	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation Does measure 3 planes of action
Active movement extent discrimination apparatus (AMEDA) Hobbs et al 28 (2010)	Lumbar AJPD	Accuracy within 0.01 mm Reliability ICC 0.63 (95% CIs 0.44, 0.77)	Active test therefore more meaningful?Weight bearing and tolerated by people with pain	Possible to reproduce clinically but questionable validity given the head is moving during the active test and so vestibular system could be contributing
Neck Proprioception testing device (NPTD) Lee et al 29 (2005)	Cervical AJPD	None stated	Active test therefore more meaningful?	Possible to reproduce clinically but questionable validity given the head is moving during the active test and so vestibular system could be contributing
Proprioceptive apparatus (no name) Cuomo et al 30 (2005)	Shoulder PJPD PMDT	None stated	Acceptable for research population and those in pain Three planes to cover more functional motion	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation Does measure 3 planes of action
“Manipulandum” Bevan et al 31 (1994) (also described by Cordo et al, 32 1994)	Elbow PJPD	None stated	Acceptable for healthy research population Fore-arm motion is semifunctional	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
“Manipulandum” Wong and Henriques 34 (2009)	Upper limb PMDD	None stated	Acceptable for healthy research population Forearm motion is multiplanar and therefore more functional	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
“Kinarm” Bhanpuri et al 33 (2012)	Elbow PJDT PMDT	Discriminating power between healthy controls and people with cerebellar dysfunction	Acceptable for research population with cerebellar dysfunction	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
“Manipulandum” Simo et al 35 (2011)	Upper limb PMDT	Discriminating power between healthy controls and people with stroke (P < .0001)	Acceptable for research population with stroke	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Shuttle Miniclinic constant-resistance device Lin et al 36 (2006)	Hip and knee PJPD	Discriminating power between elite players and amateur/novice Cite absolute error and variable error as demonstrations of validity and reliability respectively (no data)	Acceptable for research population Combined hip and knee motion is semifunctional	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Proprioceptive apparatus (no name) Friden et al 37 (1996)	Knee AJPD PMDT	Test–retest—with 1 month delay, no significant differences for any of the 3 tests (no data)	Acceptable for research population and with ACL deficiency. Knee flexion—semifunctional but in side lying	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Proprioceptive apparatus (no name) Fischer-Rasmussen and Jensen 38 (2000); Barrack et al 39 (1983); Corrigan et al 40 (1992); Co et al 41 (1993) Also Barrack et al 42 (1989) is deviser Also Pai et al 43 (1997) Also Lephart et al 44 (1996) “PTD.” Also Hurkmans et al 45 (2007)	Knee AJPD PMDT	Discriminating—Can detect difference between injured ACL and non-injured ACL knees (P < .02) Distinguishes effect of age—increased threshold with increased age and with degeneration 39 Threshold detection distinguishes injured from noninjured knee; 41 ACL tears from noninjured; 42 trained gymnasts from nontrained 44 Increased threshold with increased age and with OA 43 Intra- and interrater reliability in OA knees (ICC 0.91) 45	Acceptable for research population and with ACL deficiency	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Proprioceptive device (no name) Brindle et al 46 (2010)	Knee/ankle PMDT PMDD	None stated	Acceptable for healthy research population	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Movement detection apparatus Matre and Knardahl 47 (2003), further described by Matre et al 48 (2002)	Ankle PMDT PMDD	Established stable baseline which suggests test–retest reliability	Acceptable for research population. Ankle flexion with foot plate—closed chain semifunctional	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Passive ankle joint repositioning test Fu et al 18 (2007)	Ankle PJPD	Absolute error in passive repositioning of ankle to 5° PF gives ICC 0.84 for accuracy and repeatability	Acceptable for research population. Ankle flexion with foot plate—closed chain semifunctional but prone which is not	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Cervicocephalic kinesthesia Kristjansson et al 49 (2001) tests from multiple authors Also Cervical Joint position error (JPE) Treleaven et al 50 (2003); Treleaven et al 20 (2006) Also JPE-Torsion Chen and Treleaven 51 (2013)	Cervical AJPD	Test–retest ICCs between 0.35 and 0.90. Question usefulness of ICCs Concurrent validity correlates with reduced balance, smooth pursuit in WADS 20 Discriminates between healthy and WADS50,51	Acceptable for research population and those with WADS. Cervical motion functional with emphasis on natural head posture and motion in all planes Neck torsion test arguably discriminates between neck and vestibular signs	Not feasible clinically as complex equipment for motion capture and requires technical expertise for data analysis (feasible in specialist clinic and if used for retraining) Precise data for interpretation and feedback
Thoracolumbar proprioception Koumantakis et al 52 (2002) (original protocol by Gill and Callaghan, 19 1998)	Spinal AJPD	Reproducibility (non-LBP) ICCs 0.76-0.80 and SEM 0.91° to 1.34°; less for LBP group. Could not discriminate between LBP and non-LBP.	Acceptable for research population and those in pain Three planes to cover more functional motion	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Arm position-matching task Dukelow et al 53 (2010); Debert et al 54 (2012)	Upper extremity AJPD	Could discriminate between healthy and traumatic brain injury. Associated with dependence in ADL	Acceptable for research population Three planes to cover more functional motion	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Active Knee joint reposition sense Bullock-Saxton et al 13 (2001)	Knee AJPD	Precision—Polhemus shown to have 0.2-0.3° total error for representation of physiological joint motion76,77 Did not discriminate between Dom and Ndom leg, nor age groups except older had greater error for PWB	Acceptable for research population Attempt at more functional application by having WB and PWB conditions	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Active Knee joint reposition sense Kramer et al 55 (1997)	Knee AJPD	Reliability “modest” 0.09 to 0.65 coefficients. Could not discriminate between symptomatic and nonsymptomatic. No correlation between sitting and standing.	Acceptable for research population and those with knee pain. Attempt at more functional application by having WB and PWB conditions.	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Active knee joint reposition test Marks et al 56 (1993)	Knee AJPD	Could discriminate between healthy and OA Precision and test–retest—SEM intersession 0.33-0.95°	Acceptable for research population and those with knee pain. Attempt at more functional application by having WB and NWB conditions.	Not feasible clinically as complex equipment and requires technical expertise for data analysis Precise data for interpretation
Limb copying and reproducing tests Kaplan et al 57 (1985)	Knee AJPD	Could discriminate between older and younger women Precision—SEM for goniometer 5°	Acceptable clinically and for research participants Semifunctional	Clinically able to be performed Measurement accuracy arguable not as precise
Limb copying test As reported in various including Paqueron et al 58 (2004); Goble et al 59 (2012)	Elbow (any limb) AJPD	Could discriminate between anesthetized state and nonanesthetized state 58	Acceptable clinically and for research participants Semifunctional	Clinically able to be performed Measurement accuracy arguably not as precise
Joint position sense Reported variously in Lord et al60-62 (1991, 2000, 2002); Sturnieks et al 63 (2004); Whitney et al 64 (2005); Tiedmann et al 65 (2007); Wood et al 66 (2008). Apparatus design by De Domenico and McCloskey 67 (1987)	Lower limb AJPD	Concurrent validity—Some association with tests of postural sway, reduced stair performance (varied). Could discriminate between arthritic and not 63	Acceptable clinically and for research participants Semifunctional	Clinically able to be performed with some specialized equipment Use of “touch” on the screen arguably a confounder for proprioception
Proprioceptive assessment: Modified Thomas splint with Pearson knee flexion piece Reported variously in Warren et al 68 (1993); Barrett et al 69 (1991)	Knee PJPD	Normal knees—no difference between males/females; right/left Discriminate between age ranges (linear regression, P < .01); OA knees vs normal (P < .001); replaced vs normal (P < .02). Wearing elastic bandage increased accuracy by 40% 69 Concurrent validity—Measure of proprioception correlated with function (r = .84) and with patient satisfaction (r = .9)Test–retest: r = .82 Discriminated between ant cruciate deficient vs ant cruciate reconstruction 18 Discriminated between arthroplasty and OA (former is better); KR with retained PCL has better JPA than if PCL sacrificed 68	Acceptable clinically and for research participants Not especially functional as on mock leg	Clinically able to be performed with some specialized equipment
Rivermead Assessment of Somatosensory Perception (RASP) Winward et al 70 (2002) in Tyson et al 14 (2008)	Elbow, wrist, thumb, ankle and big toe Proprioceptive subtests PMDT PMDD	Interrater and test–retest (intra-rater) reliability for total score: Pearson correlation coefficient for both 0.92 and for individual proprioceptive tests retest scores were 0.83 (movement) and 0.50 (direction); face, content validity given since all tests drawn from traditional clinical tests; discriminated significantly between people with and without brain damage (P < .001); and concurrent validity between proprioceptive scores and Motricity Index (Spearman correlation coefficient, r = .31 and .32, both significant) and with Barthel ADL Index (r = .35 and r = .41) 70	Acceptable clinically and for research participants Semifunctional	Clinically able to be performed Use of “touch” on the limb or digit arguably a confounder for proprioception
Cumulative Somatosensory Impairment Index (CSII) Deshpande et al 71 (2010)	Lower extremity Proprioceptive subtest AJPD	Discriminates between healthy and diabetic (P = .017), peripheral arterial disease (P = .006), stroke (P = .001)	Acceptable clinically and for research participants Semifunctional	Clinically able to be performed Measurement accuracy arguably not as precise
Distal Proprioception Test (DPT) Richardson 15 (2002)	Toe, ankle PMDT PMDD	With 1 of 2 other signs (Achilles tendon reflex or vibration detection) could discriminate peripheral neuropathy; good to excellent agreement for interrater reliability (κ range 0.677-1.00) 14	Acceptable clinically and for research participants Semifunctional	Clinically able to be performed Use of “touch” on the limb or digit arguably a confounder for proprioception Probably large ceiling effect due to dichotomous outcomes (guessing)
Dual Joint Position Test (DJPT) Beckman et al 72 (2013)	Digits PJPD	Could discriminate between healthy controls and patients (MS or vasculitis) for upper and lower extremities whereas single DPT did not	Acceptable clinically and for research participants Semifunctional	Clinically able to be performed Use of “touch” on the limb or digit arguably a confounder for proprioception
Standardized sensory assessment for children Cooper et al 73 (1993)	Finger PMDT PMDD	Test–retest and interrater reliability both 100% agreement Cutoff score for TD children 5/5 73	Acceptable clinically and for research participants Semifunctional	Clinically able to be performed Use of “touch” on the limb or digit arguably a confounder for proprioception Probably large ceiling effect due to dichotomous outcomes (guessing)
Thumb finding test Smith et al 74 (1983)	Thumb/upper limb PJPD	None stated	Acceptable clinically and for research participants Semifunctional	Clinically able to be performed Use of “touch” on the limb or digit arguably a confounder for proprioception
Wrist position sense test Carey et al 75 (1996)	Wrist PJPD	None stated	Acceptable for clinical and research participants	Clinically readily able to be performed, score out of 20

Abbreviations: AJPD, active joint position detection; PJPD, passive joint position detection; PMDT, passive motion detection threshold; PMDD, passive motion direction discrimination; ICC, intraclass correlation; CI, confidence intervals; ACL, anterior cruciate ligament; OA, osteoarthritis; PF, plantar flexion; WADS, whiplash and associated disorders; LBP, low back pain; ADL, activities of daily living; Dom, dominant; Ndom, nondominant; SEM, standard error of measurement; ant, anterior; KR, knee replacement; PCL, posterior cruciate ligament; JPA, joint position awareness; MS, multiple sclerosis; TD, typically developing; WB, weight bearing; NWB, non–weight bearing.

Discussion

We have conducted a thorough search of the literature to identify the most commonly used tools for the measurement of proprioception in humans. Not surprisingly, we found a plethora of different tools, and the variations came with respect to

Measuring different proprioceptive subsenses (active versus passive position, motion detection, or direction discrimination)

Measuring different joints

The use of different types of equipment and values

Differing populations.

Given the complexity of proprioception and the multifacetted aspects of kinesthesia and position sense as identified in the introduction, it is not surprising that the tests are so seemingly disparate. By highlighting this issue we hope that future clinicians and researchers are aware of the variations in terminology and practice and that some consensus can be reached about the definition and assessment of proprioception.

It is interesting to contemplate the influence of current understandings in the physiology of proprioception on the testing of position and motion. We did not find strong indications that there was any clear incorporation or acknowledgement of the primary role of muscle spindle activity, nor was the influence of thixotrophy consistently acknowledged (controlling for muscle activity prior to testing). There was little acknowledgement of the differentiation of individual proprioceptive senses, with most tests not controlling for efferent influences (via electromyography for example). Therefore, both the muscle afferents and efferents may contribute to proprioception in the so-called passive tests. Well-described examples of test procedures exist in the experimental literature that do control for these factors (ie, thixotrophy, muscle effort, force feedback; see Gandevia et al ¹⁰ ), and we encourage clinicians to familiarize themselves with these procedures. Most tests were conducted in mid-ranges and therefore not engaging the joint receptors as limit detectors, highlighting the emphasis of tests for muscle spindle afferent activity.

In considering the active tests there are also questions to be considered. Bevan et al ³¹ reported that the joint angle may be less accurate than the angular distance travelled by the limb, especially given the latter has more functional meaning. Gandevia and Burke ⁹ originally suggested that repositioning or relocating after active positions or motions may rely on central motor programs rather than a memory of proprioceptive coordinates. Some tests (eg, Kristjansson et al ⁴⁹ ) attempted to overcome this by increasing the complexity of the active motion; however, this was for head motion and they clearly could not control for vestibular input into the relocation/repositioning response so the specificity of the test for neck proprioceptors is not clear.

Tests for the sense of force or effort are still not routinely conducted in clinical settings; therefore, they were not available to be included in the review. We are aware of clinical researchers using various means to evaluate the sense of heaviness (see Konczak et al ⁷⁸ ) but there seems to be no consistent uptake of these constructs in rehabilitation settings as yet. Clear dialogue between researchers and clinicians is required to further robust investigations into the clinical import of these constructs in rehabilitation and recovery.

Some authors did attempt to reproduce more applied tests such as weight-bearing or through more functional arcs of motion rather than anatomical single planes of action. We were unable to determine if these are any more accurate or valid beyond an element of face validity in terms of relevance to daily life. Further to this it has been questioned how functional these point-in-time measures are. First, they do not take into account the role of fatigue on proprioception. Fatigue has been reported to influence proprioception whereby exercise-induced fatigue can produce inaccuracies of proprioception, ⁶ such as errors in matching torque, ⁷⁹ movement, or position.^80,81 Furthermore, clinical fatigue (from conditions such as chronic fatigue syndrome) may be associated with kinesiophobia and other sensor-motor impairment. Second, the tests only test conscious proprioception, whereas it is clear that mostly we are reliant on subconscious proprioception in daily activities. Further issues include potential learning effects and the role of attention levels. Some measures did attempt to control for confounding alternate sensory inputs but others did not.

Some of the tests predate current proprioceptive understanding—it may be considered a limitation of this review that we did not exclude tests devised before the 1970s. However, we could not be confident that tests produced in more recent times in fact do take into account current theories; therefore, we have included all tests found and leave it to the readers to evaluate for themselves the relative validity of the tests.

We summarized the reported clinimetric properties for each identified test and found that very few had a full suite of properties evaluated and/or reported. Therefore, there is a question mark over many in terms of robustness, believability, and utility in the clinical setting. There is a clear trade-off between more research-oriented measures and those used clinically. The former measures should be more precise, accurate, and valid (particularly in terms of controlling for input other than proprioceptive); however, these techniques may be beyond the scope of a general clinical environment. In contrast, the clinically friendly tests are arguably more for “screening” given their lack of precision. We advise readers to carefully consider the constructs they wish to test, the individual joints, and with what precision. For example, if high precision is needed to evaluate the impact of interventions for lowered proprioception, then the expense of the more research-oriented tools may be warranted. Certainly it has been our experience that an easy to conduct, clinically based tool such as the distal proprioception test is not useful in clinical trials (see Lynch et al ⁸² ). However, if the use is for a clinical screen to simply identify that there is a proprioceptive impairment that may be affecting function in the clinical setting, then the RASP ⁷⁰ certainly seems to offer a useful means to standardize the conduct and reporting of such an impairment. Subsequently, it is then up to the clinician to establish an association between the sensory impairment and the loss of function. Some tests have established an overall concurrent validity with aspects of functional performance, for example, the cervico-cephalic kinasthesia tests correlate with reduced balance and smooth pursuit in people with whiplash ²⁰ ; the knee test using a modified Thomas splint correlated with function and with patient satisfaction ⁶⁹ ; and proprioceptive tests within the RASP correlated with scores for motricity and activities of daily livings. ⁷⁰

We recommend collaborations between neurophysiologists, conducting precise proprioceptive testing, and clinicians, interested in understanding the influence of impaired proprioception. This would further a more universal appreciation of the interdependent nature of the various facets of proprioception and motor function, clarify confusion around terminology, and serve to identify and standardize tests that have validity and utility for clinical populations in their rehabilitation. Facets of proprioception missing from clinical practice such as measures of sense of force, effort, motion speed, and so forth, are areas we wish to particularly highlight for further collaboration.

Conclusion

In summary, the information provided by this review should be directly useful to clinicians and researchers alike to compare and contrast their testing needs and therefore select the most appropriate tool for the job at hand. We have identified that whatever the need, proprioceptive tools are generally poorly evaluated in clinical settings and further research is required to establish reliability and validity as a starting point in the existing tests. Current understandings of proprioception from the research literature need to be applied in clinical practice to further implement evidence-based assessment and therefore rehabilitation.