Left/right judgements, Psychophysiology, Implicit motor imagery, Pain, Cortical body representation, Working body schema, Proprioceptive representation
© 2020 Wallwork et al.
This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. For attribution, the original author(s), title, publication source (PeerJ) and either DOI or URL of the article must be cited.
Cite this article
Wallwork SB, Leake HB, Peek AL, Moseley GL, Stanton TR. 2020. Implicit motor imagery performance is impaired in people with chronic, but not acute, neck pain. PeerJ 8:e8553 https://doi.org/10.7717/peerj.8553
Left/right judgement performance was analysed using the data from the second and fifth left/right judgement tasks. The first and third tasks were included as methodological controls for task features and the fourth task was included to address a separate question that was unrelated to the aims of the present study. Specifically, the first left/right judgement task (neck images) was included to familiarise participants with the task and to allow for a known learning effect ( Boonstra et al., 2012 ). The third task (identical to task 2) was included as a backup in the situation that people with pain had difficulty completing task 2 and there were significant missing data. Because there were sufficient data available for task 2, we decided (prior to analysis) to not include data from the third task to minimise potential participant fatigue that was likely to occur given it was the third repetition of neck images. The fourth task was a separate task that included contextual images (with various backgrounds and distractions). Data from the fifth left/right judgement task (hand images) were analysed as a control because the task involved implicit motor imagery of a remote body part, and the use of new images (hands vs. necks) reduced potential for fatigue.
Participants’ data were excluded if the second and fifth left/right judgement tasks were not completed in their entirety. Consistent with previous literature, data from images were excluded if the response time was less than 500 ms as this was considered too short of a time to make a judgement response and therefore would likely represent a guess ( Wallwork et al., 2013 ). Further, if the response time for eight consecutive images reached this 5 s limit (i.e., the participant timed-out) then the data from those images were excluded because it was assumed the participant was distracted or the internet/computer failed. We included all other responses that were 5 s (i.e., timed-out response) as it was assumed the participant simply took that long to respond or was unsure about that particular response. Last, data sets were excluded where participants did not provide necessary information for covariate analysis (age, gender, handedness) or group allocation (i.e., neck pain). At the completion of data processing, there were 1,404 complete participant datasets.
Participants were grouped based on: (1) duration of pain (i.e., no pain; pain for less than 3 months considered ‘acute pain’; pain for 3 months or more considered ‘chronic pain’); (2) location of pain (i.e., left-sided neck pain, right-sided neck pain, bilateral neck pain and no neck pain).
All statistics were performed using SPSS 23.0.0 (SPSS, Chicago, IL, USA). Accuracy and response time data were both tested for normality. Accuracy of responses were not normally distributed and as a result were log transformed. The log transformed accuracy values met normality criteria (as assessed by visual inspection of P–P plots and non-significant Shapiro-Wilk statistic) and were used for all analyses. Thus, for accuracy data, the analysis results were back transformed to provide group specific data (mean, 95% CI). The back transformed mean differences and their 95% CIs were not reported, because the difference between logarithms of two geometric means results in a logarithm of their ratio, not of their difference ( Bland & Altman, 1996 ). Age, gender and handedness are known to affect left/right neck judgement performance ( Wallwork et al., 2013 ); response time increases with age, is greater in females, and is greater in left-handed people, and accuracy reduces with age. Therefore, these variables were considered in all analyses and included as co-variates where appropriate. Further, a linear regression was performed using accuracy and response time to assess for a possible speed-accuracy trade-off (i.e., faster performance but incorrect response) which would suggest improper performance of the task. In all analyses, the alpha level was set at 0.05, with a Holm-Bonferroni correction used for all multiple comparisons ( Holm, 1979 ).
To determine whether neck pain and its duration is associated with impaired left/right judgement performance (Aim 1), univariate ANOVAs were conducted (one each for accuracy and response time) for neck images, with a between-subjects main effect of Pain Duration (no pain, acute pain and chronic pain) and Age as a covariate. If a significant main effect, independent t-tests were used to make specific between group comparisons. Identical analyses were completed for left/right judgement performance on hand images (control).
To determine whether the location of neck pain is associated with impaired performance for responses to left-turning and right-turning neck images (Aim 2), accuracy and response time were separately investigated using a 2 (within-subjects main effect of Side of Head Turn in Image: left-sided turning images and right-sided turning images) by 4 (between-subjects main effect of Location of Pain: no pain, left-sided pain, right-sided pain, bilateral pain) repeated measures ANOVA. Such an analysis allowed us to determine if those with bilateral pain were equally impaired for left vs. right images as well as explore whether or not there were task-based features that influenced performance, regardless of the presence or location of neck pain. Therefore, in addition, to specifically evaluate the effect of lateralised pain on performance, repeated measures ANOVAs were completed comparing only those with lateralised neck pain (left-sided neck pain vs. right-sided neck pain) for performance (accuracy and response time) on left-sided turning and right-sided turning neck images.
Given that people with neck pain could have neck pain in one location (i.e., left side neck pain), but experience pain with a movement in the opposite direction (i.e., experience left-sided neck pain when rotating their neck to the right), we ran a sensitivity analysis classifying participants into groups based on the direction of neck rotation which induced their pain. This was completed to ensure that we did not miss a potential effect of the location of neck pain (see Supplemental Information ).
A total of the 1,737 people who completed the online task, 333 were excluded due to incomplete data for the neck pain questionnaires. This resulted in a sample of 1,404 participants from 35 countries. In Test 2, 546 of the 56,160 single responses (i.e.,