Abstract
Background Proprioceptive imprecision is believed to contribute to persistent pain. Detecting imprecision in order to study or treat it remains challenging given the limitations of current tests.
Objectives The aim of this study was to determine whether proprioceptive imprecision could be detected in people with neck pain by testing their ability to identify incongruence between true head motion and a false visual reference using the Proprioception Incongruence Detection (PID) Test.
Design A cross-sectional study was conducted.
Methods Twenty-four people with neck pain and 24 matched controls repeatedly rotated to specific markers within a virtual world and indicated if their true head rotation was more or less than the rotation suggested by the visual feedback. Visual feedback was manipulated at 6 corrections, ranging from 60% of true movement to 140% of true movement. A standard repositioning error (RPE) test as undertaken for comparison.
Results Healthy controls were better able to detect incongruence between vision and true head rotation (X̅=75.6%, SD=8.5%) than people with neck pain were (X̅=69.6%, SD=12.7%). The RPE test scores were not different between groups. The PID Test score related to self-reported pain intensity but did not relate to RPE test score.
Limitations Causality cannot be established from this cross-sectional study, and further work refining the PID Test is needed for it to offer clinical utility.
Conclusions Proprioceptive precision for neck movement appears worse in people with neck pain than in those without neck pain, and the extent of the deficit appears to be related to usual pain severity. The PID Test appears to be a more sensitive test than the RPE test and is likely to be useful for assessment of proprioceptive function in research and clinical settings.
Neck pain is one of the 20 most burdensome chronic health problems worldwide,1 with evidence suggesting that treatment effectiveness is underwhelming.2,3 Several authors have speculated that imprecision in the proprioceptive system could contribute to or cause persistent pain via peripheral4–6 and central mechanisms.7–9 This speculation has resulted in widespread use of treatments aimed at improving proprioception. These treatments have been shown to improve chronic and recurrent neck pain, regardless of whether prescribed alone10,11 or in combination with other treatments,12 but compelling evidence that improvement in symptoms is mediated by improvement in proprioception is lacking.
Proprioception includes the sense of body position, movement, force, and shape. Although proprioception is generally considered to be a result of peripheral signals, the last decade of research has elucidated the critical role of central factors, such as the integrity of cortical proprioceptive maps,13 and the important role of vestibular and visual inputs in body-related perception. Proprioceptive deficits have been shown to exist in a number of persistent pain problems, including low back pain,5,6,14,15 knee osteoarthritis,4 and complex regional pain syndrome16,17 (see Lotze and Moseley18 for a relevant review).
Recently, a meta-analysis by our group confirmed that proprioceptive deficits probably exist in people with neck pain relative to controls when measured using a repositioning error (RPE) test.19 The individual studies included, however, generally had insufficient statistical power to detect differences, thus yielding null findings.20–24 These findings demonstrate that between-group differences are small relative to the high variability, both in the population and in the RPE measurement.25 This lack of test sensitivity limits the usefulness of the RPE test in investigating the role of proprioception in persistent pain states. Furthermore, the RPE test may be insufficient to capture the multiple dimensions of proprioception that can be altered in pain states. We contend that better knowledge of proprioception and its relationship to clinical states would be helpful and that techniques that allow us to gain this knowledge need to be developed.
We aimed to determine whether proprioceptive deficits could be detected in people with neck pain by testing their ability to identify incongruence between true head motion and a false visual reference, using a novel test we have called the Proprioception Incongruence Detection (PID)26 Test. We hypothesized that people with neck pain would show deficits in PID Test scores relative to controls and that scores would relate to pain severity. As secondary research goals, we aimed to assess the strength of the relationship between the PID Test and the traditional RPE test and to compare their relative sensitivities.
Method
Participants
Participants with neck pain were recruited through local physical therapy clinics and reimbursed AU$20 for participation. Participants were included if they had had neck pain for more than 8 weeks, had no overt neurological signs, and their attending physical therapist deemed them fit to tolerate repeated neck rotation. More than 8 weeks was required because the test involves repeated movement, which may be inappropriate for the acute phase of some injuries. Furthermore, pain-related brain changes that may contribute to proprioceptive imprecision have not been consistently found with time frames less than 8 weeks27 and may increase with time.28,29 All participants had pain associated with movement, as they also were involved in a concurrent study dependent on this feature.30 Age- and sex-matched healthy controls were recruited through local physical therapy clinics, the BodyinMind.org website, university campus noticeboards, and social media. In order to be considered age-matched, participants were required to be within 5 years of the associated participant with neck pain. Data were collected at The University of South Australia in August through September 2013. The sample size was calculated based on the ability to detect a between-group difference in test scores of 7% with 80% power, considering the known variability in test scores was a standard deviation of approximately 10% during pilot studies. Sample size was calculated for a single-tailed comparison in line with our clear and informed hypothesis.
Twenty-four individuals (6 men, 18 women; mean age=44 years, SD=15) with long-standing neck pain and 24 age- and sex-matched controls (mean age=45 years, SD=15) volunteered to participate in this study. The average duration of neck pain was 12 years (SD=10, range=2 months–25 years), and the pain conditions were described as primarily relating to posture or tension (n=7), degeneration (n=3), whiplash (n=5), other trauma (n=2), or scoliosis (n=1). Participants with neck pain were mildly to moderately disabled (average Neck Disability Index score=29%, SD=13%), and no participants reported neck-related dizziness.
Equipment
In order to isolate neck movement, participants sat in a well-supported chair with torso movement restricted using a seatbelt. Participants wore an Oculus Rift virtual reality head-mounted display (HMD) (Oculus VR, Menlo Park, California). The HMD displayed the virtual world and recorded head movement using on-board gyroscopes. Customized software was used to specify the factor, or rotation gain, by which “real-world movement” was translated into “virtual movement” and to map 6 different scenes to the virtual template. Participants wore white-noise–emitting headphones to counteract any environmental sounds that might inform head orientation.
PID Test
Participants repeatedly rotated their head slowly to the left or right, stopping at a marker placed within the virtual environment at 20 degrees of virtual rotation. At this point, participants were asked, “Is your true amount of head rotation more or less than the virtually simulated rotation suggested?” After verbal and written explanation of the test (see Appendix), a practice phase was initiated in which the rotation gain was set to half (2 trials) or double (2 trials) their real-world movement. Participants were required to detect the incongruence in all 4 trials to show that they understood the task before continuing. Each test consisted of 7 trials at 6 different randomly presented rotation gain settings (totaling 42 trials), taking approximately 12 minutes to complete. The 6 rotation gain settings manipulated the virtually simulated rotation to 60%, 76%, 92%, 108%, 124%, and 140% of true rotation in order to create varying degrees of incongruence and, therefore, task difficulty. After each rotation trial, the participant verbalized the direction of perceived incongruence and returned to center. The next scene and rotation gain setting were then loaded, and the task was repeated to the opposite side. The experimenter, who recorded participant responses, was blinded to the direction of the incongruence in each trial. Six different virtual scenes were used so as to prevent accommodation to the environment. The marker was placed at 20 degrees of virtual rotation so that even those participants with significantly limited movement would be able to complete the task. Each participant's score was calculated as the percentage of correct responses across the 42 trials.
Repositioning Error Test
For the repositioning error test, participants again wore the HMD and were instructed to find their neutral head position. They were then asked to turn left or right and were lightly guided to approximately 30 degrees of neck rotation by the experimenter's hands, which enabled the degree of rotation to be standardized among participants. From this point, participants were asked to refind their original position as accurately as possible. Repositioning error was operationally defined as the absolute angular difference in rotation between starting (neutral) and finishing positions and was calculated from the average of 6 trials (3 left and 3 right). This is known as the absolute error. Although there are several other ways to calculate error, they generally reduce or make little difference with regard to test sensitivity.31 Relocation to neutral was chosen over relocation to a target position, as this measure has been shown to be most sensitive to differences between patients with neck pain and healthy controls.19 During this test, a black screen was presented on the HMD to eliminate visual feedback, and participants also were instructed to close their eyes. Angles were measured using the on-board gyroscopes, which we previously confirmed had very high accuracy for angular measurements and were thus valid.30
Reliability of the PID Test
In order to establish the reliability of the PID Test, a sample of 9 healthy controls underwent repeated-measures testing. Reliability was assessed using the intraclass correlation coefficient (ICC) between repeated measures. The PID Test was shown to have high correlation among repeated measures, showing the test to have very good reliability (ICC=.74). Scores on trial 1 (77%) and trial 2 (73%) were comparable, excluding the possibility of a rapid learning effect. The participants also repeated the test using 45 degrees of virtual motion, rather than the usual 20 degrees, to ensure that this limited range of motion would not confound results. The testing confirmed that performing the test to 20 degrees had high agreement with performing the test to 45 degrees (ICC=.79).
Clinical Data
To determine whether severity of pain related to proprioception, participants were interviewed prior to testing. The variables explored were duration of symptoms, worst pain intensity, and average pain intensity experienced over the last week. Participants reported their pain using an 11-point numerical rating scale, where 0=“no pain” and 10=“the worst imaginable pain.” Neck Disability Index and Tampa Scale of Kinesiophobia (TSK) measures were administered for descriptive purposes and for exploratory analyses examining the relationship between fear of movement and PID Test results.
Data Analysis
In order to test the hypothesis that people with neck pain would be less able than healthy controls to detect incongruence in the PID Test, between-group differences were assessed using a single-tailed t test with alpha set at .05. The hypothesis that PID Test scores would relate to higher pain severity was tested using univariate regressions. The relative sensitivity of the PID Test to between-group differences was determined by observing its ability to identify between-group differences compared with that of the RPE test. The correlation between the different tests of proprioception was assessed using Pearson r. In order to check that the PID Test results were not confounded by a perceptual bias toward overperceiving magnitude of movement in the neck pain group, the number of trials where real movement was rated as more than simulated movement was compared between groups using chi-square analysis.
Role of the Funding Source
Professor Moseley is supported by a National Health and Medical Research Council (NHMRC) Principal Research Fellowship (ID 1061279). Ms Madden is supported by the Oppenheimer Memorial Trust, South Africa. This study was supported by NHMRC Grant ID 1047317. Dr Meulders is a postdoctoral researcher of the Research Foundation, Flanders, Belgium (FWO Vlaanderen) (Grant ID 12E3714N).
Results
Neck Pain and Proprioception
People with neck pain were less able to detect incongruence between visual and proprioceptive input than were healthy controls. That is, people with neck pain scored lower (X̅=69.6%, SD=12.7%) on the PID test than did healthy controls (X̅=75.6%, SD=8.5%) (t46=1.9, P=.03, d=0.55; Fig. 1). People with neck pain (X̅=3.3°, SD=1.5°) were not significantly less accurate on the RPE test than healthy controls (X̅=2.8°, SD=1.1°) (t46=1.3, P=.11, d=0.36; Fig. 2).
Percentage of correct responses for the Proprioception Incongruence Detection (PID) Test in people with neck pain and healthy controls. CI=confidence interval.
Repositioning error (RPE) test scores for people with neck pain and healthy controls. Absolute error scores are given in degrees. CI=confidence interval.
Clinical Correlates of the PID Test
The PID Test scores related to average pain (r=.38, P=.04). Secondary analyses suggested that PID Test scores also related to worst pain (r=.55, P=.004) but not to duration of pain (r=.06, P=.4). No relationship was found between the RPE test and clinical measures (P>.3 for all). The RPE test and PID Test scores did not relate when analyzed across all 48 participants (r=−.18, P=.11), nor when analyzed within the neck pain (r=−.21, P=.16) and control (r=.00, P=.50) groups.
Perceptual Bias
In order to check that differences in PID Test scores were a result of imprecision rather than bias, we compared the number of trials where real-world movement was rated as more than virtual movement with the number of trials where real-world movement was rated as less than virtual movement. We expected 50% of trials to be rated as less and 50% to be rated as more. On average, participants in the neck pain group rated 49% (SD=15%) of the trials as less and 51% (SD=15%) of the trials as more, whereas the control group rated 48% (SD=16%) of the trials as less and 52% (SD=15%) of the trials as more. Chi-square analysis showed that there was no difference between less and more ratings (χ2=0.00, P=1.0), confirming that the between-group differences were not a result of perceptual bias. No relationship was found between kinesiophobia score (TSK) and PID Test score (r=.21, P=.17), confirming that perception of magnitude was not biased by fear of movement.
Discussion
We aimed to determine whether proprioceptive imprecision could be detected in people with persistent neck pain by testing their ability to identify incongruence between true head motion and a false visual reference using the PID Test. We hypothesized that people with neck pain would show deficits in PID Test scores relative to controls and that scores would relate to pain severity. Our primary hypotheses were supported. That is, people with neck pain were less precise in detecting incongruence relative to healthy controls, and this deficit related to pain intensity.
PID Test Properties
Current tests of proprioception have fallen short of revealing group differences unless very large samples are used.19 That the PID Test was able to reveal group differences in a sample of 48 participants, therefore, represents a substantial improvement in test sensitivity. The greater sensitivity of the PID Test does not appear to be explainable by differences in test variability, given that the PID Test's reliability was comparable to that of the RPE test.32,33 Instead, the PID Test may have greater sensitivity because it includes trials of varying difficulty, making it less susceptible to floor and ceiling effects. The PID Test's greater sensitivity to between-group differences also may be a result of the specific dimensions of proprioception that are probed by the test. Furthermore, that the RPE test is ordinal and the PID Test is dichotomous and that the tests use different numbers of repetitions may bring about differences in test precision and sensitivity.
No relationship was found between the PID Test and RPE test. This finding might have been predicted on the basis of RPE data showing limited agreement among RPE test methods.33 Nonetheless, they ostensibly interrogate similar properties, so one would expect a relationship. Perhaps this finding suggests that PID Test and RPE test assess different aspects of proprioception, which is certainly a multidimensional capacity. Given that individual clinical tests are limited in their capacity to quantify the many dimensions of proprioception,34,35 the lack of relationship between the 2 tests may enable them to contribute uniquely to a barrage of tests that might more adequately quantify proprioception, in a way that no single test could. Indeed, until we can adequately quantify proprioception, our ability to test its contribution to persistent pain, motor control, and other phenomena will be limited.
Notably, the relationship between tests may be confounded by the nonproprioceptive elements inherent in each test. During the RPE test, for example, participants are required to remember a position and then replicate it. Clearly, the test also relies on short-term memory. The PID Test, on the other hand, requires real-time comparison of visual and nonvisual proprioceptive information. This comparison removes any significant memory component but might introduce a greater, or at least different, cognitive element. The PID Test also may depend on the salience of nonvisual proprioceptive input (somatosensory and vestibular) in the face of contradictory visual information, which may vary among individuals.
Sensory Imprecision and Persistent Pain
Evidence exists for imprecision or disruption in the processing of body-related information across multiple systems in people with chronic pain. The conceptual framework that has been proposed to make sense of these disruptions is that of the cortical body matrix—a network of neural loops that maintains the multisensory representation of the body and peripersonal space and subserves its regulation and protection at a behavioral and perceptual level.17 There are established principles that govern the operation of large populations of brain cells, such as the neural mass principle, which suggests that the stability of this multisensory representation would depend on the presence of highly tuned neurons and effective intracortical inhibition, among other factors.36 Deficits in sensory processing and sensory perceptions are likely to reflect disruption of these inhibitory mechanisms. This idea is supported by the correlation between clinical tests of tactile acuity and cortical changes in people with chronic pain.37–42 That cortical changes and tactile acuity also correlate with pain intensity adds support for the relevance of this finding.37–42
Compared with tactile acuity, the relationship among proprioceptive acuity, pain, and cortical mechanisms has received less attention. The impaired performance on motor imagery tasks, which rely on intact cortical proprioceptive maps, demonstrated in a number of pain states43–46 does suggest that a relationship may exist. That the perceived size of a phantom limb varies in proportion to cortical change and pain intensity47 further implies a link between disrupted proprioceptive maps and pain. Given the clear relationship between PID Test performance and pain intensity, the PID Test may be a relevant tool for investigating disruptions in proprioceptive maps.
How cortical imprecision might lead to persistent pain remains unclear. The recently proposed imprecision hypothesis suggests that imprecision leads to greater difficulty in differentiating learned signals of threat and signals of safety, in turn leading to erroneous defensive responses, including pain.8 Similarly, Zaman et al9 propose that when sensory acuity is impaired, discrimination between sensations that have become unpleasant (through fear learning) and nociceptive sensations becomes increasingly difficult, resulting in more painful perceptions. Other authors have proposed that the incongruence between motor intent and proprioception might itself cause pain,7 although evidence for this suggestion is conflicting.48,49
Limitations and Future Directions
This study, to our knowledge, represents the first work developing and examining the PID Test in a clinical population. Therefore, although the results are promising, they clearly need to be replicated. In light of recent evidence elucidating working memory and executive function deficits in people with persistent pain,50,51 we suggest caution in the interpretation of any test in which there is memory or cognitive load—which there is in both the PID Test and the RPE test. In addition, we note that the PID Test results might be influenced by the various strategies used by participants in determining whether their movement was more or less than that suggested by the false visual reference. For example, from postexperiment interviews, it seemed that some participants compared how far it felt like they rotated with how far it looked like they rotated, whereas others compared how fast it felt like they were turning with how fast it looked like they were turning. Although these 2 strategies both rely heavily on proprioception, a third strategy used by some participants may not. That is, some participants compared how fast or far it looked like they were turning with how fast or far they expected it to look like they were turning. This method would equate to sensing a difference between visual feedback and motor intention. Knowledge of these different strategies could inform future designs that attempt to further elucidate the effect of strategy used on performance. Passive movement might offer important information that removes the motor or effort component but would also add new sensory cues, such as from kinetics of the experimenter's guidance.
Further work refining the test will be needed before the PID Test can offer clinical utility—for example, categorizing performance as within or beyond normal limits, as has been done for tactile acuity.52 This issue also applies to the RPE test, and that more than double the participants would be required for the RPE test to yield sensitivity similar to that of the PID Test suggests that the widespread clinical adoption of the RPE test might have been premature.
We note that other efforts are being made to expand the repertoire of tests able to quantify sensorimotor characteristics of the neck, such as those that measure smoothness of linear motion.53–55 These tests also may offer greater sensitivity and contribute to suit of tests that can gauge the various aspects of proprioception.
Implications
Neck pain appears to be related to a lesser ability to detect when visual and nonvisual cues are incongruent during head rotation. Furthermore, this ability appears to relate to pain severity, which highlights its potential relevance. This information adds to our growing understanding of sensory imprecision in persistent pain states. The PID Test appears to gauge aspects of proprioception that are distinct to the RPE test and exhibits greater sensitivity to proprioceptive deficits. We, therefore, suggest that the PID Test may be a relevant tool for the study of proprioception.
Appendix.
Participant Instructions for the Proprioception Incongruence Detection Test
Footnotes
Dr Harvie, Ms Madden, and Professor Moseley provided concept/idea/research design. Dr Harvie, Dr Hillier, Ms Madden, Dr Meulders, and Professor Moseley provided data analysis and writing. Dr Harvie provided data collection. Dr Smith and Dr Broecker provided software design and equipment. All authors approved the manuscript for submission. Professor Moseley provided fund procurement.
Ethical approval for the study was granted through the Institutional Ethics Committee of The University of South Australia.
Professor Moseley is supported by a National Health and Medical Research Council (NHMRC) Principal Research Fellowship (ID 1061279). Ms Madden is supported by the Oppenheimer Memorial Trust, South Africa. This study was supported by NHMRC Grant ID 1047317. Dr Meulders is a postdoctoral researcher of the Research Foundation, Flanders, Belgium (FWO Vlaanderen) (Grant ID 12E3714N).
- Received April 16, 2015.
- Accepted September 13, 2015.
- © 2016 American Physical Therapy Association