The results of our study revealed very good (0.8 < ICC < 0.9) test retest reliability values for all patient groups for the visual and kinaesthetic subscales of the KVIQ-G short and long version
. Exceptions showed MS and Parkinson’s disease groups for the visual KVIQ-G subscale in both versions. ICCs for the total KVIQ-G (vis. + kin.) showed a very good overall test retest reliability for all groups.
Patients in the chronic group and MS patients produced unacceptable (below 0.6) Imaprax-G ICCs that lead to an unacceptable overall ICC according to DeVellis
. Nevertheless, the subacute, PLI, and Parkinson’s disease groups produced very good and respectable test retest reliability values.
Based on the original literature by Malouin et al. (2007) no group differences were expected
. Regardless the higher average MI vividness scores in the KVIQ-G (KVIQ-G short: 1.4-2.7 points, KVIQ-G long: 3.6-6.6 points, KVIQ-G total: 8.0-11.9 points) compared to the work of Malouin et al. our analyses computed slightly lower ICC values showing a difference from 0.01 to 0.15 compared to Malouin’s reported ICCs for CVA patients
. The higher KVIQ-G values could be explained by the educational effect of the questionnaire order applied and its resulting longer MI vividness evaluation duration. Likewise, the KVIQ-G subgroup ICC values of our Parkinson’s disease group were lower (0.01-0.14) than in the values reported for KVIQ in the work of Randhawa et al. (2010)
. Nevertheless, values were higher in the total KVIQ-G compared to values in
. At T0 patients in the MS group showed the lowest KVIQ-G scores for the visual and kinaesthetic subscales in short or long version as well for the Imaprax-G. This remained the same for T1 except for the short and long visual KVIQ-G subscale. We hypothesise that the below average scoring could be related to the longest disease duration of 14.8 years on average, compared to the other patient groups. Overall, KVIQ-G scores were higher for the visual subscale than for the kinaesthetic subscale at T0 and T1, and higher for the affected side than for the non-affected side, which corresponds to the reports of Malouin et al. and Randhawa et al.
[8, 10, 19]. Additionally, KVIQ-G 20 values for Parkinson’s disease were about 10 points below the ones reported by Randhawa and colleagues for visual and kinaesthetic subscale
. This could be due to the longer disease duration of more than 4.6 years in the current study. MS patients were the only patient group, whose KVIQ-G scores increased for all versions between 0.8 and 3.3 from T0 to T1. Scores of all other patient groups decreased from T0 to T1 for all KVIQ-G versions between 0.5 and 1.5. Several reasons could have contributed to this change. While aiming for constant test conditions, we are aware of day-to-day differences or medication dependencies of the questionnaire results. Furthermore, the construct of MI is very abstract. At T1 patients were more familiar with the topic and rated themselves more critically. We hypothesised that there will be a learning effect for all assessments and questionnaires that will be performed more than once with the same patient or healthy person. In the present study, we believed that patients did learn about the evaluation procedure and the construct of MI, which most of them did not know before. We do not believe that patients learned all movements or gestures they had to imagine, or remembered the different rating scales for KVIQ in detail, or the Imaprax video they had chosen during the first time.
Our SEM values for all patients were similar to the reported values from Malouin et al. for CVA patients in the visual and kinaesthetic subscales of the short KVIQ-G version
. Both KVIQ-G subscales revealed clearly lower SEM values for the long version.
Cronbach’s alpha values for internal consistency showed higher values (0.88 to 0.96) compared to Malouin and colleagues (2007) indicating a high item homogeneity for the construct assessment of MI vividness
Our concurrent validity testing did not result in a high correlation value as expected. We assume that the correlation result is influenced by the number of patients available for this analysis rather than differences between KVIQ and Imaprax in the construct MI vividness. KVIQ-G required patients to assume an internal perspective, whereas Imaprax-G let patients choose the MI perspective. As a result, only 19 patients did select the internal MI perspective for both questionnaires and could therefore be included in the validity analysis.
To provide further inside in the structure of the KVIQ-G a factor analysis was performed. In the statistic literature different item participant ratios for exploratory factor analysis are suggested varying between 1 vs. 3 to 1 vs. 10
. Our successfully performed analysis with a ratio of 1:3.7 confirmed the bifactorial structure of the KVIQ-G.
The MI perspective
The Imaprax-G let patients freely select a MI perspective for each of the six gestures. Frequency analyses showed that the minority of the patients, whether in a disease group, or classified for dominant or non-dominant hand affect, selected the internal perspective. A possible explanation for the phenomenon could be found in Li (2000), who observed an impaired MI of limbs in the internal, but not in the external MI perspective
. Furthermore, Kim et al. (2009) investigated the exercise-related imagery perspective in middle-age adults and reported an internal vs. external perspective ratio of 1.8
. Mulder et al. (2007) could show a slightly better MI vividness in adults over 64 years when using the external MI perspective
. The KVIQ-G requires patients to use the internal MI perspective but did not consider testing for a patient’s understanding of the MI perspective component. Therefore, we recommend determining the spontaneously selected MI perspective in patients or their ability to distinguish between both MI perspective options, before administering the KVIQ-G or any other MI vividness assessments. This could be done by using the Imaprax-G questionnaire or ask the patients what they have seen in their ‘inner picture’ after performing an MI gesture or movement example.
The parietal lobe debate
Various studies discussed the involvement of the right and left parietal lobe in MI and attributed the parietal cortex to be responsible for MI movement generation and preservation
[24–26]. After stroke aspects of this function can be affected resulting in impaired MI performance. Sack et al. (2005) found evidence that the right parietal lobe may overtake left parietal functions
. Considering the results of our LPI group no divergently KVIQ-G or Imaprax-G values could be detected. Therefore, we are suggesting to not excluding patients from MI interventions based on a lesion in the left parietal lobe only. Nevertheless, patients MI vividness evaluation should not be limited to vividness only as tested with assessments, e.g. KVIQ, or Imaprax. Further procedures, e.g. mental chronometry, mental rotation, to evaluate different aspects of patients MI vividness should be used to confirm assessment findings in general
The small sample size in the MS (N = 7) and Parkinson’s disease (N = 9) groups raised two issues in the data analysis and reported results: Firstly, the CI lower bounds dropped to negative values and thus, the concurrent validity between KVIQ-G visual and Imaprax-G visual could not be determined. Nevertheless, in the original publication by Malouin et al. a subgroup with five participants with a lower limb mobilisation was investigated
. In the current investigation, presented results included results for different subgroups and the total study population. It is believed that the chosen sample size is helpful to detect potential differences. An inference statistical analysis was avoided.
Secondly, a low number of patients (N = 19) could be included in the validity correlation analysis, due to their selection of the internal MI perspective when administering the Imaprax-G questionnaire. Further aspects should be considered when interpreting the data: Firstly, the current study did not include a control group to compare MI questionnaire values with age-matched healthy controls. Based on the findings of Malouin et al., we did not expect variations in the assessment scoring between patients and healthy controls
. Secondly, Imaprax-G and KVIQ-G were always administered in the mentioned order. We avoided a random procedure to prevent influencing patients in their MI perspective selection for Imaprax-G.