Psychometric Properties of 3 Functional Mobility Tests for People With Parkinson Disease

Reliability

The ICC values for interrater and intrarater reliability of all 3 tests were .95 or higher, and all had narrow confidence intervals. In this study, we opted for a single STS test. A previous study investigated test-retest reliability of the Five-Times Sit-to-Stand Test and reported a lower value (ICC=.76).⁴ The authors also did not explain who rated the test and how it was timed exactly, making reproducibility and integration into clinical practice difficult. We opted for a single STS test because the literature suggests that the multiple STS test is more a test of endurance, which was not the aim of our measurement.¹⁹ Furthermore, differences in ICC results could be explained by the fact that Suteerawattananon and Protas⁴ examined test-retest reliability, whereas our study evaluated intrarater reliability. Patients redoing a test can be a source of variability, leading to possibly lower ICC values. Reliability values obtained for the TUG and BM tests in this study were comparable to values obtained in the study by Morris and colleagues.⁵

We noted with interest the slightly increased mean difference and widened 95% confidence interval in Figure 3A, presenting the Bland-Altman plot for interrater agreement of the BM test. A possible explanation for this finding is that although we used a thorough standardized protocol, our agreed-upon definition for the end of movement could be further improved, as there were occasions when recording the endpoint of the BM test was difficult to assess due to participants shuffling on the edge of the bed. This subjective element of timing could have caused the discrepancies among raters. Future research could further standardize the endpoint of the BM test with an additional definition or cue such as resting with hands on the lap and coming to a complete stop.

Validity

We found moderate correlations between the STS test and part III of the UPDRS and Hoehn and Yahr stage. Although comparable for the Hoehn and Yahr stage, correlations were slightly higher and classified as “good” for the TUG and BM tests and part III of the UPDRS. We believe that these correlation coefficients indicate construct validity for our 3 functional mobility tests, as all measures have a similar underlying construct, and thus positive (significant) correlations indicate that longer timings to perform the tests are related to higher UPDRS part III scores and Hoehn and Yahr stage, with both latter measures giving higher scores to worse motor and mobility performances. Because we were unable to find studies relating the STS test with the UPDRS and Hoehn and Yahr stage, we are unable to compare our results with the previous literature.

Our results for the TUG confirm the findings in previous studies, which also showed a positive correlation between TUG timings and the UPDRS score.^20,21 Our literature search did not result in a previous study exploring the relationship between BM and UPDRS score or Hoehn and Yahr stage.

Discriminant Ability and SDD

All 3 tests discriminated significantly between the PD and control groups and between people in Hoehn and Yahr stages I and II and those in Hoehn and Yahr stages III and IV but not between nonfallers or single fallers and repeat fallers or between nonfreezers and freezers. Our results confirm the findings of previous biomechanical studies that people with PD are slower getting up from a chair compared with people who are healthy.^22–27 Our sample of people with PD was 38% slower than their healthy counterparts, which is slightly lower than but comparable to the 50% reported by Mak and Hui-Chan.²² Some authors have argued that reduced hip flexion, difficulties switching from flexion to extension, and lower hip torques related to reduced hip strength correlate with difficulty rising from a chair in people with PD.^22–27 Thompson and Medley⁷ investigated discriminant ability of the TUG for the different Hoehn and Yahr stages. They concluded that the test was able to detect differences between people in Hoehn and Yahr stages I, II, and III and people who were are healthy; thus, our results are in line with their findings. We opted to combine Hoehn and Yahr stages I and II and stages III and IV based on the relatively low number of participants in each stage, especially stage IV (n=3). Recently, Nocera et al⁸ suggested discriminative ability for the TUG in people with PD, but this suggestion was based on a different methodological approach. Their results indicated that, overall, 74% of their sample could be correctly classified, but classification of fallers (sensitivity=54%) was lower compared with classification of nonfallers (specificity=85%).⁸ Our literature search did not result in a previous study investigating discriminant ability of a BM test.

Of interest was the fact that although the timings of the 3 functional mobility tests were all relatively slower for repeat fallers in comparison with nonfallers or single fallers and for freezers in comparison with nonfreezers, the measurements did not reach significance, and thus no discriminant ability could be established for these subgroups. The fact that the STS, TUG, and BM tests are specific functional mobility measures and falls are multifactorial in nature may explain these results.¹⁴ Similarly, freezing is typically a problem that occurs within gait,²⁸ thus limiting a possible relationship with the STS and BM tests. The TUG does include walking (and turning), and indeed the results in Table 2 show the relative difference between nonfreezers and freezers is larger for the TUG in comparison with the STS and BM tests and the P value for the TUG comparison shows a trend toward significance (P=.07).

For all 3 functional mobility tests, we established applicable SDDs, which can be used in clinical practice as well as rehabilitation research but, due to the lack of examining test-retest reliability, are limited to timing of single performances in groups of patients having similar variance to our sample. Scientists as well as clinicians should continue to be aware of the difference between statistical and clinical difference. In order to consider these differences, properties such as the SDD are useful. When looking at the results in Table 2, differences in timings for all 3 tests were statistically significant between the PD and control groups and between people in Hoehn and Yahr stages I and II and those in stages III and IV, but the differences also exceed the SDD values that we reported in this study, thus further validating the discriminant ability we found in our sample. To our knowledge, this is the first study that provides such a comprehensive overview of psychometric properties for functional mobility measures in people with PD. These measures can now be used with confidence in larger-scale studies. Schenkman et al²⁹ reported previously on functional limitations and task performance in a sample of 339 people with early- and middle-stage PD. With evidence of reliability and validity for measures in an elderly population, validation is necessary in more specific populations as well, such as in people with PD, as they display specific motor deficits that might influence assessment of performance. Further results from our study, especially the SDD data, may assist in interpreting the result of clinical trials aimed at improving functional mobility in people with PD.

As mentioned in the introduction, besides established psychometric properties, outcome measures for clinical practice should have good clinical utility; they should require little time, limited cost, and no specialist equipment and training, and they can be taken into the patient's own environment.³ We believe these requirements for clinical utility are all true for the functional mobility tests we evaluated. It does not take long to ask a patient to perform a single STS, TUG, or BM test. The only equipment needed is a video camera, which also can be used for other outcome measures or for filming a home exercise program for a patient, so it is not solely of use for these tests. Furthermore, when using a video camera, the tests should only be performed once; the timing can happen later. A video camera these days is not what we would call specialist equipment, and for timing the tests, no specialist training is necessary, although we must emphasize the point that training and the use of a standardized protocol should be encouraged, even for measuring with a stopwatch the timings of a functional mobility test. Finally, these tests can easily be performed at home, which we would argue would highlight the ecological validity of these tests, as the patient can use his or her own chair or bed.

There are some limitations of our study that have to be considered. Participants were recruited through convenience; therefore, our sample may not be representative of the PD population or the elderly population. Most of our sample of people with PD was classified into the first 3 Hoehn and Yahr stages, and all of these participants were in the “on” phase of medication. Our sample did not include those who are more severely affected by PD or people in the “off” phase; thus, psychometric properties of this subgroup and this phase cannot be presented.

One should also consider our sample of men and women across our groups. The healthy group was female dominated, whereas the PD group was male dominated. Differences in build and strength (eg, leg length) may affect overall speed and could have accounted for differences between the groups. Furthermore, although the raters were blinded for study group (PD or control) when observing the videos, motor problems displayed by the PD group were easily recognizable; therefore, bias could have been introduced when observing the videos of the 3 functional mobility tests. Furthermore, one has to consider the fact that we evaluated interrater and intrarater reliability and did not incorporate test-retest agreement. The reliability tested in this study addressed measurement between and among investigators, rather than between assessments.

One must also be aware that reliability based on a single measurement is a limitation of our study; intra-participant variability is possibly more important in determining whether a participant changed with an intervention than is the variability in measurement when viewing performance from a video recording. Generalizability of our reliability values to non–video-recorded performance also should be considered with caution. Finally, one could suggest that video recordings are less practical in a clinical setting compared with direct observations, but we believe that video recordings are more beneficial because they can be stored for comparison between performances over time or to discuss performance together with the patient. Furthermore, from a standardization point of view, the same performance can be viewed repeatedly, if necessary, which again is more advantageous than asking the patient to perform the test again.